hydrogen and helium traces in type ib-c supernovae

A&A 450, 305–330 (2006)DOI: 10.1051/0004-6361:20054366c© ESO 2006

Astronomy&

Astrophysics

Hydrogen and helium traces in type Ib-c supernovae

A. Elmhamdi1,2,3, I. J. Danziger3, D. Branch4, B. Leibundgut5, E. Baron4, and R. P. Kirshner6

1 ICTP-International Center for Theoretical Physics, Strada Costiera 11, 34014 Trieste, Italye-mail: [email protected]; [email protected]

2 INAF-Osservatorio Astronomico di Callurania, via Mentore Maggini, 64100 Teramo, Italy3 INAF-Osservatorio Astronomico di Trieste, via G.B.Tiepolo 11, 34131 Trieste, Italy4 Department of Physics and Astronomy, University of Oklahoma, Norman, OK 73019, USA5 European Southern Observatory, Karl-Schwarzschild-Strasse 2. 85478, Garching, Germany6 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Received 17 October 2005 / Accepted 29 December 2005

ABSTRACT

Aims. To investigate the spectroscopic properties of a selected optical photospheric spectra of core collapse supernovae (CCSNe). Specialattention is devoted to traces of hydrogen at early phases. The impact on the physics and nature of their progenitors is emphasized.Methods. The CCSNe-sample spectra are analyzed with the parameterized supernova synthetic spectrum code “SYNOW” adopting somesimplifying approximations.Results. The generated spectra are found to match the observed ones reasonably well, including a list of only 23 candidate ions. Guided bySN Ib 1990I, the observed trough near 6300 Å is attributed to Hα in almost all type Ib events, although in some objects it becomes too weak tobe discernible, especially at later phases. Alternative line identifications are discussed. Differences in the way hydrogen manifests its presencewithin CCSNe are highlighted. In type Ib SNe, the Hα contrast velocity (i.e. line velocity minus the photospheric velocity) seems to increasewith time at early epochs, reaching values as high as 8000 km s−1 around 15−20 days after maximum and then remains almost constant. Thederived photospheric velocities, indicate a lower velocity for type II SNe 1987A and 1999em as compared to SN Ic 1994I and SN IIb 1993J,while type Ib events display a somewhat larger variation. The scatter, around day 20, is measured to be ∼5000 km s−1. Following two simpleapproaches, rough estimates of ejecta and hydrogen masses are given. A mass of hydrogen of approximately 0.02 M� is obtained for SN 1990I,while SNe 1983N and 2000H ejected ∼0.008 M� and ∼0.08 M� of hydrogen, respectively. SN 1993J has a higher hydrogen mass, ∼0.7 M�with a large uncertainty. A low mass and thin hydrogen layer with very high ejection velocities above the helium shell, is thus the most likelyscenario for type Ib SNe. Some interesting and curious issues relating to oxygen lines suggest future investigations.

Key words. stars: supernovae: general – line: identification – line: profiles

1. Introduction

Stripped-envelope SNe, namely type Ib (helium-rich) andtype Ic (helium-poor), being hydrogen deficient objects, are un-doubtedly amongst the most mysterious SN classes. Recentlyefforts have started to understand the nature of these objectsthrough studies of samples (Matheson et al. 2001; Branch et al.2002).

However the rarity of cases with well sampled observa-tions, photometry and spectra, hampers a more direct infer-ence of the physical situation behind the explosions and hence aclear view of the progenitor nature. Nevertheless, the discoveryof metamorphosing events as SN 1987K and SN 1993J (rec-ognized as SNe IIb), that evolve from type II to type Ib-c asthey age, together with the similarity of the environments inwhich they occur have linked type Ib-c SNe to a core-collapsescenario in massive stars. At present, indeed, the most widely

accepted models relate type Ib-c SNe to both relatively lowmass progenitors within the context of close binary system evo-lution (i.e. mass-loss as consequence of mass transfer) and mas-sive stars that have undergone significant mass-loss due to awind (i.e. Wolf-Rayet stars). So far observations have not dis-criminated between the two scenarios, although recently, usingHST data, a high spatial resolution search for the progenitorof the type Ic SN 2004gt in a wide wavelength range from thefar UV to the near IR has suggested that the event might re-sult from an evolved Wolf-Rayet star, although the observationscould not constrain models invoking less massive progenitorsin binary systems (Gal-Yam et al. 2005).

Photometrically, the lack of significant hydrogen in theouter layers of SNe Ib-c probably inhibits the most importantcharacteristic of type II SNe light curves, namely the plateauphase resulting from the hydrogen recombination wave.

Article published by EDP Sciences and available at http://www.edpsciences.org/aa or http://dx.doi.org/10.1051/0004-6361:20054366

http://www.edpsciences.org/aa

http://dx.doi.org/10.1051/0004-6361:20054366

306 A. Elmhamdi et al.: CCSNe spectra

At late phases, the steeper decline rate, compared to the“56Co to 56Fe” decay slope, is indicative of significant γ-rayescape as a result of the low mass ejecta in this class of objects(Clocchiatti & Wheeler 1997); there are rare exceptions wherethe late slope approaches the full trapping rate (e.g. SN Ib1984L; Schlegel et al. 1989).

Spectroscopically, a clear separation scheme within thestripped-envelope SNe subclasses is still lacking. Part of theproblem is the absence of meaningful statistics. A direct clas-sification was earlier proposed by Harkness et al. (1987) on thebasis of He I strengths in the photospheric optical spectra oftype Ib SNe. Wheeler et al. (1994), however, have claimed thepresence of He I 10 830 Å in SN Ic 1990W, although He I lineswere not noted in the optical region. The authors presentedthe idea of adopting, instead, the OI 7773 Å line as a distin-guishing feature. The absorption seems stronger in type Ic thanin Ib SNe. Matheson et al. (2001) came to the same conclusionwhen analyzing a sample of Ib and Ic events. In addition, it hasbeen argued that in type Ib SNe He I lines 5876 Å and 7065 Ågradually grow in strength with respect to the one at 6678 Å(Matheson et al. 2001). SN 1998bw (GRB980425) is anotherexample of classification as type Ic, but where IR lines of He Ihave been clearly identified (Patat et al. 2001).

Branch et al. (2002) have presented a relation between thevelocity at the photosphere, measured using synthetic-spectrumanalysis, and the time since maximum light for a sample oftype Ib SNe. The relation is well fitted by a power-law, indicat-ing a fairly homogeneous behaviour. Interestingly, SN Ib 1990Idoes not follow this trend (Elmhamdi et al. 2004).

An important issue concerns features in the 6000−6500 Åregion of the early spectra in Ib-c SNe. Deng et al. (2000), whenanalyzing spectra of SN Ib 1999dn, have attributed the absorp-tion feature seen at 6300 Å around maximum brightness to Hαwhich later on disappears or is overwhelmed by C II 6580 Åwhen Hα optical depth decreases as a consequence of the enve-lope expansion. The possibility of the presence of C II 6580 Åin Ib spectra has been earlier suggested by Harkness et al.(1987). Alternatively, the 6300 Å absorption was identified tobe due to Ne I 6402 Å in SN Ib 1991D (Benetti et al. 2002).Ne I lines and Hα had already been proposed to account for thedeep absorption in SN Ib 1954A by Branch (1972).

The presence, or not, of hydrogen and/or helium intype Ib-c, with the possibility of quantifying the amount, is ofgreat importance in identifying the progenitor stars that maygive rise to these classes of objects.

The papers by Matheson et al. (2001) and Branch et al.(2002) have provided an impetus towards an advanced under-standing of SNe Ib and Ic. The present work may be regardedas a continuation of those efforts. However in our comparativestudy of early spectra of type Ib SNe we include representa-tives of all the various types of CCSNe, namely type IIb, type Icand type II. This is established by means of synthetic spectragenerated with the parametrized SN synthetic-spectrum code“SYNOW”.

Our main goal was to understand the similarities and differ-ences among SNe Ib objects in the available sample, and to alsocompare to properties of the wider CCSNe family. The paper isorganized as follows. First the analysis method and parameters

are illustrated in Sect. 2. Data description is briefly given andthe best fit synthetic spectra are presented and compared withthe observed ones in Sect. 3. This is done separately for eachindividual object among representatives of CCSNe classes.Section 4 presents two methods to obtain spectroscopic massestimates. The complete results will be discussed in detail andconclusions will be drawn in Sect. 5.

2. Fitting procedure: SYNOW code

For the purpose of our analysis we make use of the parameter-ized supernova synthetic spectrum code “SYNOW”. The codeassumes spherical symmetry, homologous expansion: “v ∝ r”velocity-law and resonant scattering line formation above asharp photosphere, emitting a blackbody continuum. SYNOWtreats line formation and line blending (i.e. multiple scattering)within the Sobolev approximation (Jeffery 1990; Fisher 2000;Branch 2001), and reads lines from the Kurucz 42-millionatomic line list (Kurucz 1993). It is an LTE code in only onerespect: LTE excitation for the relative strengths of the lines ofan individual ion, but it does not assume LTE ionization.

SYNOW provides a number of free fitting parameters. Themost important are: 1. (τref ), the optical depth of the strongestline, in the optical region, of the introduced ion. The line iscalled “the reference line”. The optical depths of the other linesof the same ion are thereafter determined assuming Boltzmannequilibrium. To decide which ion to introduce in the synthe-sis procedure we rely on the work by Hatano et al. (1999) asa starting point. The authors have presented the variation withtemperature of LTE−Sobolev line optical depths of 45 individ-ual candidate ions that might be encountered in supernova en-velopes for six different compositions. 2. (Tbb): the underly-ing blackbody continuum temperature. We did not attach highphysical importance to this parameter although the galactic ex-tinctions for each individual supernova, as reported by Schlegelet al. (1998), is taken into account. Independently of the presentstudy, estimating the total reddening to the event, both thatcaused by foreground dust in the Milky Way and that caused bydust in the host galaxy, is a crucial point in supernovae study,especially when using them as cosmological probes. 3. (vphot):the velocity at the photosphere, estimated from the match withFe II lines. Restrictions on the velocity interval within whichan ion is introduced is possible. A maximum outer boundaryvelocity of line-forming-region of 5 × 104 km s−1 is adoptedin the present analysis. When assigning a minimum ion veloc-ity, “vmin”, greater than “vphot”, the ion is said to be “detached”from the photosphere, and consequently has a “non-zero” op-tical depth only starting at “vmin”. In SYNOW the profile ofa detached line has a flat-topped emission, and the absorptionminimum is blueshifted by the detachment velocity. An unde-tached line has a rounded emission peak. A slightly detachedline has a flat top but only over a small wavelength interval.However whenever one talks about detached lines one needs tokeep in mind that we are using a somewhat unrealistic “τv”, onethat has a discontinuity in it. Real supernovae spectra probablydo not have sharp discontinuities.

The radial dependence of the line optical depths can be cho-sen to be either exponential with an e-folding velocity “ve” as

A. Elmhamdi et al.: CCSNe spectra 307

Table 1. The candidate reference lines of CCSNe phtospheric-phasespectra in the optical region (shown in increasing wavelength order).

Ion Rest Wavelength (Å)

Ca II 3934

Ni II 4067

Co II 4161

Mn II 4205

Ca I 4227

Cr II 4242

Sc II 4247

C II 4267

Mg II 4481

Ti II 4550

Ba II 4554

Fe II 5018

Mg I 5184

He I 5876

Na I 5890

Si II 6347

Ne I 6402

H I 6563

[O II] 7321

O I 7772

Si I 7944

N I 8680

C I 9095

free parameter (i.e. τ ∝ exp (−v/ve)), or a power-law with anindex “n” (i.e. τ ∝ v−n). In the present work a power-law pro-file is adopted with an index n = 8, although for some objectswe will discuss the possibility of an exponential profile.

When fitting our CCSNe sample spectra we have testedmany combinations of fitting parameters. Only the best gener-ated synthetic spectra are displayed in the following and com-pared to the observed ones. We introduce the parameter “con-trast velocity”, defined as the line minimum velocity minus tothe photospheric one: “vcont(line) = vmin(line)− vphot”. We showas well the behaviour of a similar parameter, defined instead asa ratio, i.e. “vratio

cont (line) = vmin(line)/vphot”.

Although the computations are made under the purely res-onant scattering assumption, we find that the photospheric-optical spectra of CCSNe are fitted well usually requiringonly 23 candidate ions or fewer. Table 1 lists the candidate ref-erence lines sufficent to reproduce the observed features in op-tical spectra of the CCSNe sample. It is important to note thatsince at early phases line formation takes place in high velocitylayers causing severe line blending, it is better in some com-plicated cases to analyze line identifications in reverse chrono-logical order (i.e. starting with later phase spectra followed byearlier ones). All the analyzed spectra have been transformedto the rest frame of their host galaxies.

3. Data description and analysis

Our selected CCSNe sample consists of 20 objects – 16 of themare type Ib, 2 are type Ic, 1 is type IIb and 1 is type II SN.Some of the spectra presented in this work were gatheredwith the 60′′ Telescope and the “MMT” on Mount Hopkins.They are part of the supernova monitoring program of theCenter for Astrophysics (PI R.P. Kirshner). Data for SN 1996aqare taken from the Padova-Asiago supernova database. Use ismade as well of the “SUSPECT”1 supernovae spectral archive.Descriptive data regarding the sample events are listed inTable 2 (i.e. host galaxy, recession velocity and phases). Foreach individual event, the table summarizes as well the mostimportant fitting parameters. The last column indicates thenumber of ions we find responsible for determining the bestfit spectra (details described in the following). Throughout thepresent work, however, we do not provide details concern-ing observations of individual objects. We focus, instead, onsynthetic spectra fitting procedures, constraints, problems andwhat we may learn about the CCSNe physical situation.

3.1. Type Ib SNe: the sample

∗ SN 1990I:We start with SN 1990I since it can be considered as one

of the better observed objects among type Ib-c SNe in terms ofquality and sampling. The event has been exploited both pho-tometrically and spectroscopically giving constraints on ejecta,oxygen and nickel masses and energy estimates (Elmhamdiet al. 2004). The supernova seems to follow a velocity trenddifferent from the pattern shown by the 10 SNe Ib sample ofBranch et al. (2002). Here we investigate this peculiarity bymeans of synthetic spectra fits.

The observed spectrum at maximum light, shown in Fig. 1,is compared to a synthetic spectrum (SSp) that has a veloc-ity at the photosphere vphot = 12 000 km s−1 and a blackbodycontinuum temperature Tbb = 14 000 K. The SSp containslines of He I, Fe II, Sc II, Mg II, O I, Ca II and Hα. Thelines of Sc II, with τ(Sc II) ∼ 0.4, have been introduced tohelp the fit of the absorption feature around 6000 Å and red-ward the trough attributed to He I 5876 Å. Absorption troughsof P-Cygni He I lines at 5876 Å, 6678 Å and 7056 Å areevident, although their relative strengths cannot be simulta-neously fitted within the LTE approach in the SSp. We willface this limitation each time we analyze and fit He I lines. Amore precise analysis requires NLTE treatment as He I linesmay be non-thermally excited by the decay products of 56Niand 56Co (Lucy 1991). Apart from He I and Hα lines, the re-maining lines have non-zero optical depths starting at the pho-tospheric velocity (i.e. they are undetached). The SYNOW pa-rameters required to account for the He I lines are τ(He I) ∼ 2.9and vmin(He I) = 14 000 km s−1.

The absorption minimum near 6250 Å is well fitted by Hα,with vmin(Hα) = 16 000 km s−1 and assigned a moderate op-tical depth of 0.6 (Fig. 1). However, we tested other plausiblealternative identifications that were introduced in the literature

1 http://bruford.nhn.ou.edu/suspect/index1.html


Table 2. Main data and fitting parameters of the CCSNe sample spectra.

Supernova Host Galaxy Vrecession Phases Vphot Tbb1 τ(H) τ(He I) Vmin(H) Vmin(He I) Nbr(ions)

(Type) (km s−1) (days) (km s−1) (K) (km s−1) (km s−1)

SN1990I (Ib) NGC 4650A 2902 max 12 000 14 000 0.6 2.9 16 000 14 000 7

12 10 000 5500 0.6 1.85 15 500 13 000 7

21 9500 5400 0.26 0.94 15 000 12 000 7

SN1983N (Ib) NGC 5236 513 max 11 000 8000 0.34 2.5 15 000 Vphot 8

10 7000 5000 0.55 3.4 12 000 9000 11

SN1984L (Ib) NGC 991 1534 8 8000 8000 0.65 2.5 15 000 10 000 5

32 5000 5600 0.2 6.7 12 500 7000 5

SN1987M2 (Ic) NGC 2715 1339 20 7000 4500 − 0.6 − 9000 7

SN1988L (Ib/c) NGC 5480 1856 20 8000 9000 0.7 1 16 000 11 000 9

SN1990B (Ib/c) NGC 4568 2320 8 8500 5400 0.5 0.55 13 000 10 000 7

30 7000 4000 − 0.8 − 10 000 9

SN1991ar (Ib) IC 49 4562 28 8000 4400 0.32 10 16 000 Vphot 10

48 7000 4800 0.4 4.8 16 000 8000 10

60 6000 4600 0.28 4.4 14 000 7000 12

SN1991D (Ib) PGC 84 044 12 500 21 4600 7000 0.46 1.8 12 000 6000 11

SN1991L (Ib/c) M 7-34-134 9050 31 5000 6000 0.34 2 13 000 Vphot 12

SN1993J (IIb) M 81 35 16 9000 7800 21 0.6 10 000 Vphot 10

24 8000 7000 25 1.5 9500 Vphot 10

59 6000 5000 2 30 8000 Vphot 11

SN1994I (Ic) NGC 5194 461 max 12 000 7000 − − − − 9

SN1996aq (Ib/c) NGC 5584 1602 2 9000 9000 0.12 0.35 15 000 12 000 8

24 4600 5000 − 0.25 − 7000 11

SN1997dc (Ib) NGC 7678 3480 28 5000 4200 0.2 10 13 000 7000 10

SN1998dt (Ib) NGC 945 4580 8 9000 5600 0.3 4 17 000 11 000 6

33 9000 5000 0.2 10 17 000 Vphot 9

SN1998T (Ib) NGC 3690 3080 21 6000 5400 0.34 1.15 10 500 9000 9

42 5800 5200 − 0.75 − 8000 9

SN1999di (Ib) NGC 776 4920 21 7000 4800 3 12.6 12 000 8500 8

45 6000 5200 1.6 10 12 000 7500 9

52 6000 5000 1 10 12 000 7000 9

SN1999dn (Ib) NGC 7714 4920 −10 14 000 6600 1.5 2 18 000 Vphot 5

max 10 000 6000 1 5 14 000 11 000 6

10 7000 5800 0.24 1.9 15 000 9000 10

38 6000 5400 0.34 14.5 14 000 7000 10

SN1999ex (Ib) IC 5179 3498 4 10 000 5800 1.45 2.35 18 000 11 000 7

13 7000 5600 1.6 3.4 12 000 8000 12

SN1999em (IIP) NGC 1637 717 −6 10 000 10 000 15 0.6 Vphot Vphot 3

25 4600 6400 21 − 5600 − 13

SN2000H (Ib) IC 454 3894 max 11 000 9000 5.25 2 13 000 Vphot 9

5 8000 6500 2.5 2 13 000 9000 9

19 6000 4600 2.5 5 13 000 8000 9

30 5000 6000 1.45 3.4 13 000 7000 11

47 5000 4400 0.5 1 12 000 7000 91 The events are corrected for the galactic reddening effect (Schelegel et al. 1998). Additional host galaxy extinction is corrected for only intwo objects, namely SNe 1990I and 1999em.2 The reported data corresponds to the low He I velocity case (see Sect. 3.2).


6000 6500

C II

Fig. 1. The SN 1990I spectrum, near maximum, compared with the“SYNOW” SSp (thin line). Lines that are responsible for the mostconspicuous features are also shown. The region around the 6250 Åtrough is zoomed in the window.

to account for similar features seen in type Ib SNe, namelySi II 6355Å, Ne I 6402 Å and C II 6580 Å lines. UndetachedNe I lines were quite tempting in SN Ib 1991D, and workedas well as Hα having a contrast velocity of vcont(Hα) =7000 km s−1 (Benetti et al. 2002). For SN 1990I, instead,vcont(Hα) is only 4000 km s−1 which means that undetachedNe I line is excluded because it is too blue to fit the observedfeature. Si II 6355 Å is obviously also too blue. C II 6580 Åremains then a plausible alternative for Hα, since its rest wave-length is slightly redder than Hα and therefore has its con-trast velocity as a free parameter. In the window of Fig. 1,we show synthetic spectra for both cases (Hα and C II). C II,with vmin(CII) = 17 000 km s−1 and τ(CII) ∼ 0.003, providesa fit as good as Hα. But an acceptable C II 6580 Å line wouldrequire an additional velocity of about 820 km s−1 comparedto Hα (∼18 Å difference). C II 6580 Å has been proposedin early spectra of SN Ib 1999dn (Deng et al. 2000). Indeed,at maximum light and day 14, SSp fits of SN 1999dn in-cluded C II with vcont(CII) = 7000 and 1000 km s−1 respec-tively, while He I lines were formed starting at the photosphere(i.e. vcont(He I) = 0 km s−1) (Deng et al. 2000). For SN 1990I,the situation is almost similar: He I in the SSp shown in Fig. 1has vcont(He I) = 2000 km s−1 while vcont(CII) = 5000 km s−1.This means that the C II lines would need to be formed abovethe helium layer, which would be surprising, if not unphysical,in type Ib SNe.

SN 1990I, with its particular velocity structure, presentsgood evidence for an Hα feature. However since it is not ab-solutely clearcut that Hα is always responsible for the absorp-tion at about 6300 Å seen in type Ib SNe, the presence of theHβ Balmer line would be of great support. Unfortunately theoptical depth sufficient to fit the Hα trough is so small that

Fig. 2. SN 1990I “SYNOW” fits of the observed 12 day and 21 dayspectra. Conspicuous line features are indicated.

the other Balmer features are too weak to be unambiguouslydetectable (τ(Hα) is about 7 times greater than τ(Hβ)).

Based on our investigation, the ions that generally mightbe encountered in shaping the 6000−6500 Å wavelength rangeare: Hα, Ne I, C II, Si II, Sc II, Ca I, He I, Fe II, Si I and Ba II.Here, we propose the following criteria in view of the method-ology for deciding what lines should be adopted (especially forthe 6300 Å feature):

1. when Fe II lines are very strong, they could produce atrough that might be sufficient to fit the 6300 Å feature.The depth of that Fe II feature is controlled by means ofFe II lines at 4924, 5018 and 5169 Å;

2. undetached Sc II, Si I, Ca I and Ba II can be introduced es-pecially to fit the slope at 6000−6300 Å wavelength range.They should not, however, introduce unwanted features inthe rest of the spectrum;

3. undetached Ne I 6402 Å line is rejected once its feature istoo blue to fit the 6300 Å trough or/and when the otherNe I lines clearly introduce various unwanted features.Similar reasoning applies to the Si II 6355 Å line;

4. with its contrast velocity as a free parameter, C II 6580 Åline could be a candidate for the 6300 Å trough; neverthe-less it is ruled out once it exceeds the He I contrast velocity.

In Fig. 2, the observed spectra at 12 days (top panel)and 21 days (bottom panel) are compared to the synthetic oneswith (vphot = 10 000 km s−1; Tbb = 5500 K) and (vphot =

9500 km s−1; Tbb = 5400 K), respectively. Synthetic line pro-file features are labeled by the designation of the ion whose linegives rise to the feature. He I is still detached from the photo-sphere with vcont(He I) = 3000 km s−1 at 12d and vcont(He I) =2500 km s−1 at 21 days. The He I 7065 Å seems to increasein strength relative to the He I lines at 5876 Å and 6678 Å,


Fig. 3. Synthetic “SYNOW” fit of the 21 days observed spectrum, withonly lines of SiII (top), NeI (middle) and CII (bottom).

indicating the non-thermal excitation effects are changing butstill existent. Two interesting points emerge from our SSp fit:on the one hand, even though the reference line Ca II 3933 Å,with τ(Ca II) = 120 at both 12 days and 21 days, producesa good match to the observed one, the Ca II infrared triplet(8542, 8662, 8498 Å) has a clear deficit especially in the emis-sion component of its P-Cygni profile. On the other hand, theobserved O I 7773 Å is deep. An optical depth of 2 is in factimposed to reproduce the deep trough in the SSp at 21 days.The O I 7773 Å lines are believed to have relatively greaterstrength in type Ic SNe compared to Ib objects (Wheeler et al.1994; Matheson et al. 2001). This is presumably because fora “bare” C/O envelope of type Ic, one would expect oxygenlines to be more prominent relative to Ib case where an intacthelium, and possibly some hydrogen, could tend to dilute theC/O core.

Matheson et al. (2001) have defined a parameter called“Fractional Line Depth” of the line through absorptionminimum relative to the continuum flux. Mean valuesof 0.27 (±0.11) and 0.38 (±0.091) were found to representtype Ib and type Ic respectively. For the case of SN 1990I, wemeasure a value of about 0.45, indicating that the SN was pecu-liar in this respect. One explanation for this abnormal behaviourof SN 1990I might be related to the amount of oxygen. In factElmhamdi et al. (2004) have argued for a possible high oxygenmass (∼0.7−1.35 M�) relative to other Ib objects (∼0.3 M� inSNe 1984L, 1985F and 1996N).

As far as the 6000−6500 Å wavelength region is concerned,we find that Fe II, He I and Hα are sufficient to reproduce theoverall shape in the SSp at both 12 days and 21 days. We have,however, tested the Ne I, Si II and C II possibilities as can-didates for the 6250 Å trough. Panels in Fig. 3 are similar tothe bottom of Fig. 2, but with only Si II(top), Ne I(middle)and C II(bottom) lines. All three ions are formed starting at thephotosphere (vcont = 0 km s−1). Both Si II and Ne I lines areruled out by means of criterion number 3 above. C II is as well

Fig. 4. SSp fit of SN 1983N compared to the observed spectra at max-imum light (upper panel) and at 10 days (lower panel). Conspicuousline features are shown.

rejected because of criterion 4. We therefore consider Hα to bethe most likely explanation in SN 1990I.

∗ SN 1983N:The observed spectra at maximum light and at ∼10 days

are shown in Fig. 4, together with the corresponding syn-thetic spectra. The SSp at maximum brightness has vphot =

11 000 km s−1 and a blackbody continuum temperature Tbb =

8000 K, while the one at 10 days has vphot = 7000 km s−1 andTbb = 5000 K. Ne I and C II are rejected because of the cri-teria 3 and 4, while undetached Si II helps in fitting the ab-sorption blueward the 6250 Å trough. Hα with τ ∼0.34 ac-counts nicely for the 6250 Å absorption at maximum. He Ihas τ ∼ 2.5 and vcont = 0 km s−1, while SSp at 10 days hasvcont(He I) = 2000 km s−1 and vcont(Hα) = 5000 km s1. Na I lineis added in the 10 days SSp to help the fit of the emission com-ponent peaked around 5900 Å.

∗ SN 1984L:Figure 5 compares the observed spectrum at ∼8 days with

an SSp that has vphot = 8000 km s−1 and Tbb = 8000 K. Ions thatare responsible for the most conspicuous absorption featuresare indicated. Interestingly, He I lines with vcont = 2000 km s−1

and τ = 2.5, provide a good match at 5876 Å, 6678 Åand 7056 Å. That may indicate that departures from LTE, dueto non-thermal effects, are not severe for this type Ib object. Asimilar result obtains for the 32 day spectrum i.e. (vcont(He I) =2000 km s−1 and τ = 6.7; top panel of Fig. 6). The correspond-ing SSp in Fig. 6 has vphot = 5000 km s−1 and Tbb = 5600 K.

The 32 days He I profiles show rounded emission com-ponents that cannot be matched by a power-law SSp. As wehave already mentioned, a detached line in a power-law as-sumption has a discontinuity in its optical depth (i.e. non-zero


Fig. 5. SSp fit of SN 1984L compared to the observed spectrum atday 8. Conspicuous line features are shown.

optical depth only above the vmin(line)). When a profile retainsa rounded P-Cygni emission component even if its vmin(line) isgreater than vphot, this might indicate two components of opti-cal depth rather than only one above the photosphere. In suchcases better fits could be obtained by having gradually decreas-ing optical depth below the detached velocity, instead of a dis-continuity. In the present version of SYNOW, only an e-foldingassumption of the optical depth allows a two-component treat-ment of a given line. We switch then to the exponential caseadopting ve = 3000 (see Sect. 2). For He I lines and belowvmin = 7000 km s−1 we introduce a second component withnegative ve (ve = −2000 km s−1), such that “τ” is continuous atthe detachment velocity (i.e. at 7000 km s−1). In this case andfor similar situations, the line should not be defined as “de-tached”. Instead, it has a maximum value of “τ” that is notat the photosphere as is normal for an undetached line. Thebottom panel in Fig. 6 demonstrates that the two-componentoptical depth reflects more probably the real situation of theHe I lines in SN 1984L. The He I line fits seem better relativeto the power-law case, although blueward 5500 Å we obtainan inferior match. In the 8 days SSp, Si II is too blue to ac-count for the 6250 Å. We checked the undetached Ne I 6402 Åline possibility, assigning it an optical depth of 1.5. It gives aprofile broader than the observed one. In addition the cover-age in wavelength of the spectrum hampers a check whetherNe I lines produce unwanted features longward of 7000 Å.Ne I lines would need to be non-thermally excited as the casefor He I lines in type Ib (Lucy 1991; Swartz 1993). With anoptical depth of 1.5, Ne I would have a departure coefficientfrom LTE of about 15 (Hatano et al. 1999). Additionally, thelow non-thermal effects seen in He I lines of SN 1984L (seeabove) may argue against Ne I identification. C II is also re-jected since it will require a vcont(CII) = 8000 km s−1 while atthis phase vcont(He I) = 2000 km s−1 (criterion 4). Hα remains

Fig. 6. SSp fit of SN 1984L compared to the observed spectrum atday 32. The lower panel shows the e-folding optical depth possibility(see discussion). Conspicuous line features are shown.

Fig. 7. SSp fit of SN 1988L compared to the observed spectrum atday 20. Conspicuous line features are shown.

then a plausible candidate. In Fig. 5, indeed, τ(Hα) = 0.65 andvcont(Hα) = 7000 km s−1 provide a good fit, while at 32 dayswe use τ(Hα) = 0.2 and vcont(Hα) = 7500 km s−1 (Fig. 6).

∗ SN 1988L:Figure 7 compares the observed spectrum at ∼20 d with

an SSp that has vphot = 8000 km s−1 and Tbb = 9000 K. Theobserved spectrum has been smoothed with a box size of 5.Narrow emission lines due to H II regions are present in the


6000 6500

Fig. 8. SSp fit of SN 1990B compared to the observed spectrumaround day 8. Lines that are responsible for the most conspicuous fea-tures are also shown. The region around the 6300 Å trough is zoomedin the window.

spectrum. The object has been discussed and classified as atype Ib SN (Filippenko 1988; Kidger 1988).

The strong and broad P-Cygni profile at ∼5550 Å is wellmatched in our SSp by a blend of Na I and He I lines. TheHe I 5876 Å accounts for the blue edge of the broad trough,having vcont(He I) = 3000 km s−1 and τ(He I) = 1, whileundetached Na I has τ(NaI) = 3. Ba II lines together withCa I lines have been introduced in the SSp and help the fitblueward ∼6000 Å. Undetached Si II 6355 Å is too blue toaccount for the trough seen at ∼6250 Å. We adopt Hα with anoptical depth of 0.7 and vcont(Hα) = 8000 km s−1. UndetachedC II 6580 Å, τ(CII) = 0.006, contributes together with He Iblueward 6500 Å (Fig. 7). SN 1988L may be regarded as anintermediate Ib/c object rather than a typical type Ib.

∗ SN 1990B:An extensive study of SN 1990B observations has been pre-

sented by Clocchiatti (2001). The authors pointed out the redcharacter of the object. This fact is supported in the presentanalysis, see below, by means of the estimated continuum tem-peratures that seem much lower than values at similar phasesfor the other type Ib-c sample objects.

The spectrum at day 8 with the best fit SSp is presented inFig. 8. The SSp has vphot = 8500 km s−1 and Tbb = 5400 K.Ions that are responsible for the most conspicuous featuresare indicated. The fit parameters have been modified manytimes in order to investigate line identification possibilities.The strong P-Cygni profile around 5700 Å is a blend of un-detached Na I and detached He I 5876 Å (vcont(He I) =1500 km s−1 and τ(He I) = 0.55). The expelled helium, withvcont(He I) = 1500 km s−1, accounts nicely for the weak ab-sorption near 6500 Å. Undetached Si II 6355 Å is considered

5500 6000

HeINaI

Fig. 9. SSp fit of SN 1990B compared to the observed spectrumaround day 30. Lines of conspicuous features are shown. The regionaround the 5700 Å trough is zoomed in the window. Na I D andHe I 5876 Å individual contributions are also displayed.

to fit the weak absorption at ∼6200 Å. The Hα line,vcont (Hα) = 4500 km s−1 and τ(Hα) = 0.5, is introduced to ac-count for the broad observed absorption trough around 6300 Å.The window in Fig. 8 displays a zoom of the 6300 Å region,together with the C II fit possibility (τ(CII) = 0.0025). The fitlooks as good as Hα, however it would need to be expelledwith 3000 km s−1 more than the He I lines (i.e. vcont(CII)=4500 km s−1), and hence is ruled out (criterion 4).

The second spectrum, around day 30, is compared to anSSp that has vphot = 7000 km s−1 and Tbb = 4000 K (Fig. 9).The presence of He I 5876 Å is supported by two facts, firstby the wide nature of the absorption trough that cannot becaused only by the undetached Na I D. Second, by the emer-gence of a “bump” around 5760 Å, in the transition sloperedward of the deep absorption trough at ∼5700 Å. A closerview of the “He I+Na I” region is provided in the windowof Fig. 9. The window demonstrates the contribution of bothNa I D and He I 5876 Å separately. The match is good andreproduces nicely the total feature seen in the observed pro-file. This tends therefore to confirm the presence of heliumin the ejecta of this object, even though it does not exhibit afull and clear set of helium absorption lines. He I lines havevcont(He I) = 3000 km s−1 and τ(He I) = 0.8. On the other hand,Fe II lines with τ(FeII) = 12, provide good fit in the blue partof the spectrum. In addition Fe II contributes significantly inthe absorption at ∼6320 Å. The presence of Hα is therefore notneeded since the Fe II contribution is sufficient. Neverthelessif Hα were present it would have a vcont(Hα) = 4000 km s−1.

Based on its spectroscopic properties, especially the weakHe I lines, SN 1990B was re-classified as a type Ic object(Clocchiatti et al. 2001). It was first classified as type Ib SN(Kirshner 1990). Clocchiatti et al. (2001) presented an


Fig. 10. The V-band light curve of SN 1990B compared to a sampleof Ib-c “V-light curves” (See text for details).

extensive set of well-sampled photometry. In Fig. 10, weshow the absolute V−light curve of SN 1990B compared toother SNe Ib-c. The comparison highlights the similarity withtype IIb 1993J, SN Ib 1990I and SN Ib 1991D. The maxi-mum brightness is comparable to those of SNe 1990 I, 1987Mand 1994I and intermediate between the bright SNe 1999bwand 1991D and the faint SN 1983I. Moreover the figure revealsa similar decline rate, from maximum to reach the exponen-tial decay, of SNe 1990B, 1990I, 1993J, 1991D and 1998bwand indicates similar peak-to-tail contrast, whereas SNe 1983I,1987M and 1994I display narrower peak widths and greaterpeak-to-tail contrast. The events with steeper decline rates (i.e.narrow widths) and greater peak-to-tail contrast are classifiedas type Ic SNe, while SNe 1990I and 1991D are type Ib events.This fact may point to a photometric behaviour closer to type Ibrather than type Ic. We regard, therefore, SN 1990B as an in-termediate type Ib/c event.

∗ SN 1991ar:Figure 11 displays the available spectra of SN 1991ar

at 3 different phases, namely 28 days, 48 days and 60 days.The first spectrum is compared to an SSp that has vphot =

8000 km s−1 and Tbb = 4400 K. The broad P-Cygni profilewith the minimum trough centered at ∼5720 Å is well fittedby the “He I+Na I D” blend. Although the spectrum coversonly the 4000−7000 Å wavelength range, it shows distinctlythe He I series, indicating a type Ib nature. The He I lines areundetached and have τ(He I) = 10. One particularity of thisspectrum is the flat behaviour of the ∼6070−6450 Å range. Wechecked different ion combinations, and the best fit to that zoneis reached including lines of Fe II, Si II and Hα. This latter hasvcont(Hα) = 8000 km s−1 and τ(Hα) = 0.32.

For the 48 day spectrum, the match is good with the SSphaving vphot = 7000 km s−1 and Tbb = 4800 K (Fig. 11;

Fig. 11. SSp fit of SN 1991ar compared to the observed spectraaround 4 weeks (top panel), 48 days (middle panel) and at 60 days(lower panel). Lines that are responsible for the most conspicuous fea-tures are reported.

middle panel). The He I lines are now slightly detached withvcont(He I) = 1000 km s−1 and τ(He I) = 4.8 Lines of O I andMg II both contribute to the observed features around 7600 Åand 9030 Å (τ(OI) = 3.5 and τ(MgII) = 5), while theCa II IR triplet is produced by τ(CaII) = 500. The weak fea-ture labeled as being due to Hα corresponds to vcont(Hα) =9000 km s−1 and τ(Hα) = 0.4.

The bottom panel in Fig. 11 shows the 60 day spectrumcompared to an SSp that has vphot = 6000 km s−1 and Tbb =

4600 K. The ∼6080−6500 Å range loses the flat behaviour seenin the first spectra and a round emission profiles start to form.The match in that part of the spectrum with the SSp is accept-able. The profile is mainly due to lines of Fe II, Si II, C II,Hα and Ca I. The He I fit corresponds to vcont(He I) =1000 km s−1 and τ(He I) = 4.4, while Hα has vcont(Hα) =8000 km s−1 and τ(Hα) = 0.28. Note that the two weak troughsdue to the Sc II lines at 5527 and 5661 Å appear near 5500 Å inall 3 spectra. Another notable feature is the deficiency, in flux,of the observed spectra, in the middle and bottom figures, rela-tive to the synthetic ones especially redward of ∼6500 Å. Thisis possibly because the supernova already enters a transitionphase to the nebular epoch.

Surely the evidence of Hα should not be taken as definite,although the corresponding weak absorptions in the SSp nicelymatches the observed features.

∗ SN 1991D:SN Ib 1991D data has been presented and studied by

Benetti et al. (2002). The object, with its narrow features, seemshaving lower velocities compared to other type Ib events at sim-ilar phases. When analyzing the ∼10 day spectrum by means of


Fig. 12. SSp fit of SN 1991D compared to the observed spectraaround 3 weeks. Lines that are responsible for the most conspicuousfeatures are reported. The top panel illustrates the Ne I possibility,while Hα case is displayed in the lower panel.

SYNOW code, the authors obtained an improved match intro-ducing undetached Ne I lines with respect to the Hα SSp fit.Figure 12 illustrates this possibility for the 21 days spectrum.The displayed SSp has vphot = 4600 km s−1 and Tbb = 7000 K.The He I lines are evident through our SSp (vcont(He I) =1400 km s−1 and τ(He I) = 1.8). In addition, the Na I D fea-ture (τ(NaI) = 4) contributes to the He I 5876 Å broad P-Cygniprofile. The presence of Na I lines is consistent with the goodfit around 8100 Å. The most conspicuous line absorption fea-tures are indicated in the figure. In the top panel of Fig. 12,undetached Ne I lines are included in the SSp (τ(NeI) = 2).The match in the whole spectrum is quit good. In the bottompanel the Hα possibility is tested. Except for adding Ba II linesin order to help the fit at ∼6100 Å, the other ion parameters arekept unchanged. The fit to the observed absorption near 6300 Åwith Hα, vcont(Hα) = 7400 km s−1 and τ(Hα) = 0.46, is slightlybetter compared to the Ne I case (top panel). However the SSp,in the bottom panel, does not account for the observed featuresnear 6630 Å and 6840 Å as does NeI lines in the SSp of thetop panel. Ne I remains hence a strong candidate in this type Ibobject.

It is worth noting here that at a similar phase, i.e. ∼21 days,SN Ib 1990I had a photospheric velocity of 9500 km s−1.

∗ SN 1991L:We study a spectrum dated about one month after max-

imum (Fig. 13). The spectrum has been smoothed with abox size of 3. The classification of this event based only onthe shown spectrum is very tentative. Indeed the spectrumshows a short wavelength part reminiscent of type Ia spec-tra. However, nebular spectra of SN 1991L have been found todisplay emission features that are normally found in type Ib/c

FeII

Fig. 13. SSp fit of SN 1991L compared to the observed spectra around1 month. Lines that are responsible for the most conspicuous featuresare reported. Top panel illustrates the Ne I possibility, while Hα caseis displayed in the lower panel.

events (Gomez 2002). The illustrated best fit SSp has vphot =

5000 km s−1 and Tbb = 6000 K. The He I lines are not so obvi-ous, although an undetached He I 5876 Å line, τ(He I) = 2, pro-vides a good fit to the observed absorption trough at ∼5880 Å.Na I D could contribute to that feature, although it would pro-vide an absorption component slightly redshifted. We regardSN 1991L as a transition type Ib/c object.

Similarly to SN 1991D, we tested the Hα and Ne I possi-bilities as a candidates for the feature near 6290 Å. Figure 13shows both cases. The included Ne I lines in the upper panelhave τ(He I) = 1, while in the lower panel Hα has vcont(Hα) =8000 km s−1 and τ(Hα) = 0.34. Some features due to Ne I linesappear similar to the observed ones, which are not accountedfor the Hα case (lower panel; Fig. 13). Compared to the lowerpanel, Ne I lines also improve the fit of the broad emissioncomponent of the He I 5876 Å P-Cygni profile. Ne I remainstherefore an alternative possibility to the Hα identification.

∗ SN 1997dc:The best SSp fit, vphot = 5000 km s−1 and Tbb = 4200 K,

is displayed in Fig. 14 together with the observed spectrum(∼4 weeks since maximum). On the one hand, the match withHe I series, vcont(He I) = 2000 km s−1 and τ(He I) = 2, isevident, confirming the classification as a type Ib event. Onthe other hand and in order to obtain an improved fit to theHe I 5876 Å P-Cygni profile, undetached Na I D with τ = 5is needed. The identification of Na I is consistent with the ab-sorption profile seen near 8090 Å. Ca II IR triplet is unusu-ally strong and the match to the synthetic one is rather poor.Sc II with τ = 2 is responsible for absorption features blue-ward of the He I 5876Å, and contributes as well, with Fe II,near 6160 Å.


6000 6500

Fig. 14. SSp fit of SN 1997dc compared to the observed spectrum atabout 4 weeks. Lines of conspicuous features are shown. The regionaround the 6300 Å weak trough is zoomed in the window (see text).

A zoomed view of the 5600−6600 Å region is shown inthe window of Fig. 14. The adopted SSp fits well the observedfeatures. Various ion combinations have been tested, especiallyto explain the weak absorption at ∼6290 Å. The best fit thatwould not introduce unwanted features in the rest of the spec-trum and as well be in agreement with our previous criteria(Sect. 3.1) is attributed to Hα having vcont(Hα) = 8000 km s−1

and τ(Hα) = 0.2.

∗ SN 1998dt:Spectra at 2 phases, ∼8 days and ∼33 days, are shown in

Fig. 15 and compared with the best fitting synthetic spectra.The SSp in the upper panel corresponds to vphot = 9000 km s−1

and Tbb = 5600 K, while for ∼33 days, the SSp has vphot =

9000 km s−1 and Tbb = 5000 K (lower panel). The SSp nicelymatch the features in the observed spectra. The narrow emis-sion near 6565 Å is due to Hα from H II region.

The reference Ca II line (i.e. at 3933 Å) in the upper panelhas an optical depth of 500. The corresponding Ca II H&Kand IR Ca II triplet (8542, 8662, 8498 Å) SSp both fit well theobserved broad P-Cygni profiles. For the 33 day spectrum how-ever, the Ca II infrared triplet (τ = 500) presents a deficit withrespect to the observed profile even though for the absorptionpart the fit is still acceptable. The O I 7773 Å with τ = 0.5 ac-counts for most of the observed feature at both phases (top andlower panels).

The He I lines are clearly evident for both phases, with themain difference that while at ∼8 days He I lines are detached(vcont(He I) = 2000 km s−1 and τ(He I) = 4), they have non-zero optical depths starting at the photosphere for the ∼33 dayspectrum (vcont(He I) = 0 km s−1 and τ(He I) = 10). Note herethat for both phases He I 5876 Å is sufficient to fit the observed

Fig. 15. SSp fit of SN 1998dt compared to the observed spectra at8 days (upper panel) and at 33 days (lower panel). Lines of conspicu-ous features are indicated.

P-Cygni profile. There is no need to include Na ID line. Theslope redward of the He I 5876 Å emission component is nicelyaccounted for by lines of Fe II, Ca I and Hα at ∼8 days, whilefor the 33 day spectrum we introduce, in addition, lines of Sc II(τ = 2). The presence of Sc II is supported by the double ab-sorption features near 5390 and 5500 Å.

The best SSp fit of the observed 8 day spectrum includesonly 6 elements, namely Fe II, He I, Ca I, Ca II, O I and H I.Hα, with vcont = 8000 km s−1 and τ = 0.3, accounts for theweak absorption near 6200 Å. A similar but weaker absorptionis seen in the 33 day spectrum as well. We fit it with Hα thathas vcont = 8000 km s−1 and τ = 0.2. It is clear that even thoughthe fit is convincing, the presence of Hα is not definite. Thislimitation occurs whenever we deal with absorption featuresthat are not so deep with respect to the continuum.

∗ SN 1998T:Spectra of SN 1998T at two different phases, ∼21 days

and 42 days, are displayed in Fig. 16 and compared to syn-thetic spectra having vphot = 6000 km s−1 and Tbb = 5400 K(upper panel), and vphot = 5800 km s−1 and Tbb = 5200 K(lower panel). The spectra are highly contaminated by weakemission features from an H II region. The absorption troughscaused by He I series are prominent, indicating a familiartype Ib appearance. The He I line best fit at 21 days corre-sponds to vcont(He I) = 3000 km s−1 and τ(He I) = 1.15, whileat 42 days vcont(He I) = 2200 km s−1 and τ(He I) = 0.75.The optical depth of the undetached Ca II is 120 (upper panel)and 140 (lower panel). The corresponding features, i.e. of Ca II,fit well the observed ones although an excess in the emis-sion P-Cygni components is visible, as is the case in most ad-vanced photospheric spectra. C I, with τ = 0.3, helps fit the


Fig. 16. SSp fit of SN 1998T compared to the observed spectra at 3weeks (upper panel) and at 42 days (lower panel). Lines of conspicu-ous features are indicated.

absorption feature near 8930 Å. Fe II lines are unusually weakfor SN 1998T. Indeed, optical depths of 1 and 5 are respectivelyadopted for our best synthetic fits at ∼21 days and ∼42 days.

We checked different combinations for the 6000−6500 Åwavelength range, testing the possible candidates and keepingin mind our departure criteria (Sect. 3.1). For the 21 day spec-trum, the best fit includes Si I(τ = 0.03), Si II(τ = 0.6), Fe IIand Hα. This latter has vcont (Hα) = 4500 km s−1 and τ(Hα) =0.34. When testing Ne I lines one obtains an improved fit blue-ward of the He I 7065 Å feature, but then unwanted profilesin the 6000−6500 Å region would be introduced. At low tem-peratures, an Si I identification is plausible in type Ib objectswithin LTE assumption with low optical depths (Hatano et al.1999). Si II however, is expected (in LTE) to have an opticaldepth of about 10 at temperatures similar to that for the 21 dayspectrum (helium-rich composition; Hatano et al. 1999). Thiswould mean that we might have mis-identified the absorptionfeature near 6250 Å, or we have a departure from LTE in theSi II line. Note that for type II SNe, and at similar temperatures,Si II has an optical depth as low as 0.1 (hydrogen-rich compo-sition; Hatano et al. 1999). The remaining possibility is thatthe 6250 Å is due to Hα rather than Si II. In this case hydrogenwould have vcont(Hα) = 9000 km s−1, but then the absorptionnear 6340 Å remains unaccounted for. To fit this latter with C II,we need then to impose a velocity of about vcont = 6000 km s−1,which means that carbon is expelled 3000 km s−1 greater thanhelium, contradicting our criteria (Sect. 3.1). The most accept-able combination is then the one illustrated in the top panelof Fig. 16.

The situation for the 42 day spectrum is more difficult asthe spectrum is noisier. Our best fit includes Si I(τ = 0.025),Si II(τ = 0.4), Fe II and C II (τ = 2.3 × 10−4). We pre-fer C II, having in this case similar vcont as He I, rather than

Fig. 17. SSp fit of SN 1999di compared to the observed spectra at21 days (upper panel) and at 45 days (lower panel). Lines of conspic-uous features are indicated.

adopting Hα. First, because hydrogen would need then to beexpelled at velocities lower than helium, and second becausewe remain consistent with our criteria. We propose that itwas Hα at 21 days which was then overwhelmed by C II lateras the photosphere recedes because of the envelope expansion.

∗ SN 1999di:In Fig. 17, the observed spectra of SN 1999di at ∼21 days

and 45 days are compared to the computed synthetic spectra.The spectra are synthesized adopting vphot = 7000 km s−1 andTbb = 4800 K (upper panel) and vphot = 6000 km s−1 andTbb = 5200 K (lower panel). The match is good in the overallspectral shape. The fit to the He I lines is obvious. The param-eters used are vcont(He I) = 1500 km s−1 and τ(He I) = 12.6 forthe 21 day spectrum, while at 45 days the He I optical depthdiminishes to τ(He I) = 10, keeping the same contrast ve-locity (i.e. 1500 km s−1). For the 21 day spectrum the opticalHe I lines (i.e. at 5876, 6678 and 7065 Å) are simultaneouslywell accounted for by their corresponding absorption troughsin the SSp. Even the He I 7281 Å line accounts for an observedabsorption. At day 45, however, the SN displays a shallowerHe I 6678 Å compared to the He I lines at 5876 Å and 7065 Å.At both phases, Na ID 5893 Å is blended with He I 5876 Å. Theabsorption near 8030 Å is also accounted for by lines of Na I.The features blueward of the broad “He I+Na ID” P-Cygni pro-file are attributed to Sc II lines (τ = 2 for both epochs). Sc IIalso helps the fit redward of the emission component of theblended “He I+Na ID” P-Cygni profile.

The O I 7773 Å line, with τ = 1, accounts for most of theobserved feature. Note however that the line of Mg II at 7890 Åcontributes in some cases to the red edge of the O I 7773 Åline. In the present case this contribution is not so relevant.


In the upper panel, the Ca II IR triplet, with an optical depthof 700, fits very well the observed profile. At day 45, thefit is under-estimated especially for the emission part, whichmay indicate a transition to the nebular phase. The reproducedCa II H&K line is also strong in both synthetic spectra. Theissue of the 6300 Å feature is particular in this object, becauseit is very deep and different from features seen in “normal”type Ib SNe. We have tested various identifications for the deepfeature. The best fit that would be consistent with our criteriaattributes the trough to Hα with vcont(Hα) = 5000 km s−1 andτ(Hα) = 3 (upper panel), and vcont(Hα) = 6000 km s−1 andτ(Hα) = 1.6. With these parameters and in addition to the al-most perfect Hα fit, the SSp accounts as well for the notch2

seen near 4665 Å, assigning it to Hβ. The Hα trough is rem-iniscent of what is seen in type II and IIb objects, with thedifference that in these latter types Hα displays a completeP-Cygni profile (i.e. with both conspicuous emission and ab-sorption components). This is understandable within the con-text of “detachment” concept. Indeed, Branch et al.(2002) haveshown that an Hα P-Cygni profile loses its obvious emissioncomponent when it is highly detached. If correct, we wouldthen expect an increasing value of vcont(Hα) as we go fromtype II to IIb to Ib SNe. This point will be emphasized anddiscussed at the end of the paper.

∗ SN 1999dn:Two spectra of SN 1999dn, observed at 10 and 38 days af-

ter maximum light, are compared in Fig. 18 to synthetic spectrathat have vphot = 7000 km s−1 and Tbb = 5800 K (upper panel)and vphot = 6000 km s−1 and Tbb = 5400 K (lower panel).As shown in both panels, the prominent lines of He I 5876,6678 and 7065 Å are clearly accounted for, indicating a typicaltype Ib nature. The fit to the He I reference line correspondsto vcont(He I) = 2000 km s−1 and τ(He I) = 1.9 at day 10,and vcont(He I) = 1000 km s−1 and τ(He I) = 14.5 at day 38.A spectral analysis of SN 1999dn has been also presented byDeng et al. (2000). The authors discussed line identificationsfor 3 photospheric spectra (at −10, 0 and 14 days from maxi-mum). The observed trough at ∼6250 Å was first blended withHe I 6678 Å for the −10 day spectrum, while around maximumlight the two troughs become distinctly isolated, giving rise to adouble absorption profile (Fig. 2; Deng et al. 2000), very simi-lar to what is seen in SN 1990I (Fig. 1), although the two eventsevolve afterwards in different ways, with the ∼6250 Å featurebeing less deep in SN 1999dn than in SN 1990I later on. Theauthors found it difficult to attribute the minimum near 6250 Åto Si II 6355 Å. They argued that it was Hα first, before max-imum, that becomes blended and overwhelmed by C II line inlater spectra. For the two early spectra, i.e. at −10 days andmaximum, C II 6580 Å provides a fit as good as Hα, howeverC II 6580 Å is assigned a minimum velocity much higher thanthe one attributed to He I. Highly detached C II lines are surelyimprobable in this class of event. Even in their late spectrum(i.e. around 14 days), the C II has a value of vcont higher thanthat of He I.

2 The word “notch” referes to a narrow and weak absorbtionfeature.

6000 65006000 6500

Fig. 18. SSp fit of SN 1999dn compared to the observed spectra at10 days (upper panel) and at 38 days (lower panel). Lines of con-spicuous features are indicated. The region around the 6300 Å weaktrough is zoomed in the window (see text).

We analyzed various possibilities among line candidates forour 10 and 38 day spectra. The wavelength range of interest,∼5800−6600 Å, is zoomed in the window of Fig. 18. We findthat the best fits which would reproduce the weak absorptionfeatures and not contradict the fit criteria, include lines of He I,Fe II, Ca I and Hα. This latter accounts nicely for the absorptionnear 6260 Å with vcont(Hα) = 8000 km s−1 and τ(Hα) = 0.24(at 10 days), and vcont(Hα) = 8000 km s−1 and τ(Hα) = 0.34(at 38 days).

∗ SN 1999ex:A good set of early data for SN 1999ex, spectra and pho-

tometry, have been presented by Hamuy et al. (2002). Thequality of the photometry, starting well before maximum light,offers a unique possibility of looking at the early behaviourof light curves. Indeed, the early “dip” seen in the “U” and“B” light curves is interpreted as being due to the shock break-out, supporting the present belief that type Ib-c SNe are theoutcome of core collapse in massive stars rather than thermonu-clear disruption of white dwarfs (Hamuy et al. 2002; Stritzingeret al. 2002). Because of weak optical He I lines, the objectwas classified as an intermediate case between Ib and Ic SNe.The evident trough at ∼6250 Å was attributed to Si II 6355 Å(Fig. 4; Hamuy et al. 2002). The authors also presented threeinfra-red spectra.

We analyze, by means of synthetic spectra, line identifi-cations in this interesting object. The two observed spectra,around 4 and 13 days, are compared with our best fit SSp thathave vphot = 10 000 km s−1 and Tbb = 5800 K (Fig. 19) andvphot = 7000 km s−1 and Tbb = 5600 K (Fig. 22). In Fig. 19,we combined the IR spectrum with the optical one in order to


Fig. 19. SSp fit of SN 1999ex compared to the observed spectrum at4 days. The observed IR spectrum is combined with the optical one inorder to allow the He I 10 830 Å identification (see discussion). Thelines of conspicuous features are reported.

check the consistency of He I identification. In fact, at 4 days,assigning the following parameters: vcont(He I) = 1000 km s−1

and τ(He I) = 2.35, we obtain a good match to the observedHe I profiles. On the one hand, a strong support for the presenceof He I lines comes from the good fit to the IR He I 10 830 Å.Furthermore, in Fig. 20 the fit is extended beyond 1 mi-cron. The good match to other He I-IR lines, adopting equalvmin(He I) and τ(He I) as in the optical part of the spectrum, isclear and provides unambiguous evidence for the presence ofhelium in this event. The OI 7773 Å line with τ = 0.5, on theother hand, is found to be not so strong and deep as is the casein most type Ic events (Matheson et al. 2001). These two factspoint to a type Ib nature, more than a type Ic class.

In the lower panel of Fig. 19 we test the exponential case(see details in the “SN 1984L” part; Sect. 3.1). We assign twocomponents to the He I lines: one above vmin = 11 000 km s−1

with ve = 3000 km s−1 and one below vmin = 11 000 km s−1

with negative ve (ve = −2000 km s−1), such that τ is continu-ous at the detachment velocity (i.e. at 11 000 km s−1). The fitis slightly improved for He I lines at 5876 Å and 10 830 Å andalso for the IR triplet Ca II.

As far as the trough around 6250 Å is concerned, wechecked the Si II identification attributed by Hamuy et al.(2002). We test as well the Ne I possibility. The closerview in Fig. 21 illustrates the two possibilities. In the upperpanel undetached Ne I lines can fit the feature, however theywould introduce unwanted features in the rest of the spectrum.Furthermore, because of the depth of the observed trough,a good match with Ne I 6402 Å would require an opticaldepth of 4, indicating a departure factor from “LTE” of ∼40(Hatano et al. 1999), highly improbable although non-thermal

Fig. 20. The 4 day IR spectrum of SN 1999ex compared with syn-thetic spectra that contain only lines of He I. The prominent featuresattributed to He I lines are indicated.

Fig. 21. The 4 day spectrum of SN 1999ex is compared with syntheticspectra that contain only lines of Ne I (top panel) and Si II (bottompanel).

excitation of Ne I needs to be more thoroughly investigated(i.e. through NLTE and hydrodynamic SSp codes). The Si IIpossibility is shown in the bottom of Fig. 21. The undetachedSi II 6355 Å, with τ = 2.5, is rather blue to account for thefeature. We note here that if one adopts C II 6580 Å, thenone needs to assign it a very high velocity, about 8000 km s−1

higher than the one of He I. The most likely identification there-fore remains Hα. In fact the best fit in Fig. 19 is achievedusing vcont(Hα) = 8000 km s−1 and τ(Hα) = 1.45, while forday 13 Hα has vcont(Hα) = 5000 km s−1 and τ(Hα) = 1.6(Fig. 22). The match at both phases is quite good. The otherlines responsible for shaping the spectra are labeled in the fig-ures. In addition to the convincing match to lines of He I,


Fig. 22. SSp fit of SN 1999ex compared to the observed spectrumat day 13. Lines of conspicuous features are shown.

IR-Ca II, O I, Fe II we note the presence of a weak absorptionnear 4680 Å that is well accounted for by Hβ in our 13 daysSSp (Fig. 22). This can be taken as support for the presence ofhydrogen expelled at high velocities in this supernova.

It would have been of great interest to look at later spectraof this object, in order to see how the trough assigned to Hαwould evolve with time.

∗ SN 2000H:The type Ib SN 2000H is considered one of the more in-

teresting Ib-c objects, especially with a strong and deep troughnear 6300 Å (Benetti et al. 2000; Branch et al. 2002). Herewe analyze three spectra at different epochs, namely at maxi-mum light, 19 days and 30 days. Figure 23 compares the ob-served spectrum around maximum with the resulting best fitSSp which has vphot = 11 000 km s−1 and Tbb = 10 000 K. Theoverall match is quite good. Undetached He I lines, with τ = 2,provide a good fit with the He I 5876 Å, while features assignedto He I lines 6678 Å and 7065 Å are very weak. The absorptiontrough near 6280 Å is exceptionally broad and cannot be ac-counted for only by Hα. We identify two possibilities for whichwe obtain a broad absorption in agreement with the observedfeature. The upper panel in Fig. 23 shows the “Hα+Ne I” com-bination, while in the lower panel the “Hα+Si II” combinationis illustrated. In both panels a closer view of the 6280 Å regionis displayed in the window. In the first case, undetached NeIwith τ = 1.5, a contribution blueward of He I 7065 Å improvesthe fit with the observed feature, however the He I 5876 Å emis-sion component is under-estimated. Note that the narrow ab-sorption in the emission peak is attributed to Na ID interstellarline originating in the parent galaxy. In the lower panel, unde-tached Si II (τ = 3) improves the match with the observedabsorption trough without altering the resulting SSp at the

6000 6500

6000 6500

Fig. 23. SSp fit of SN 2000H compared to the observed spectrumat maximum light. Lines of conspicuous features are indicated. Thetop panel shows the “Hα+Ne I” possibility for the broad featurenear 6300 Å. A corresponding zoomed view is displayed in the win-dow. Similarly, the bottom panel shows the “Hα+Si II” case.

Fig. 24. SSp fit of SN 2000H compared to the observed spectrum at19 days. The lower panel shows the e-folding optical depth possibility(see discussion). Conspicuous line features are shown.

emission part of the He I 5876 Å P-Cygni profile. As a result webelieve that the“Hα+Si II” combination is the most probable.

Figures 24 and 25 compare spectra at 19 and 30 dayswith their corresponding best fit SSp. These have vphot =

6000 km s−1 and Tbb = 4600 K (Fig. 24) and vphot =

5000 km s−1 and Tbb = 6000 K (Fig. 25). The two spectra aresimilar in having narrower features than the maximum light


Fig. 25. SSp fit of SN 2000H compared to the observed spectrumat 30 days. The lower panel shows the e-folding optical depth pos-sibility. Conspicuous line features are indicated.

spectrum, resulting from observing at smaller radii where ex-pulsion velocities are lower. Lines of Ca II, [O II], O I, Mg IIare seen to develop. Introducing Sc II lines helps to form a fea-ture blueward of the strong He I 5876 Å P-Cygni profile. Adistinct feature appears redward of the IR Ca II profile, moreevident in the 30 day spectrum, and is accounted for by C I(τ = 0.2 at 19 days and τ = 0.8 at 30 days). The noticeablechange compared to the spectrum at maximum is the develop-ment of the He I 6678 Å and 7065 Å troughs. The He I ref-erence line has vcont(He I) = 2000 km s−1 and τ(He I) = 5at day 19, while at day 30 the corresponding He I parametersare vcont(He I) = 2000 km s−1 and τ(He I) = 3.4.

Similarly to what is seen in the case of SN 1984L, theHe I profiles retain rounded emission components that can-not be matched by a power-law SSp. We tried to improve theHe I fits by switching to the e-folding assumption for the opti-cal depth, introducing a continuous two-component behaviourof the He I optical depth: one above vmin = 8000 km s−1 withve = 3000 km s−1 and a second component with negative ve(ve = −2000 km s−1), such that τ is continuous at the detach-ment velocity (i.e. at 8000 km s−1 for the 19 day spectrumand 7000 km s−1 for the 30 days one). It is important to re-call here that the line should not be said to be “detached”. It hashowever a maximum value of τ that is not at the photosphere asis ordinarily the case for an undetached line. The bottom panelsin Figs. 24 and 25 illustrate the resulting synthetic spectra in thee-folding cases. The fit is somewhat improved compared to thepower-law case. The noticeable improvements concern in par-ticular the He I features and the infra-red Ca II profile. We notethe good match to the ∼6250 Å trough with Hα. The best fitis achieved using vcont(Hα) = 7000 km s−1 and τ(Hα) = 2.5

Fig. 26. SSp fit of SN 1987M compared to the observed spectrum at20 days. The upper panel shows the high He I velocity case, whilethe bottom panel displays the low He I velocity case (see discussion).Conspicuous line features are shown.

at day 19 (Fig. 24), while for day 30 Hα has vcont(Hα) =8000 km s−1 and τ(Hα) = 1.45 (Fig. 25). No alternative iden-tification to Hα, that would be logically acceptable, has beenfound. In addition, the Hα identification is strongly supportedby the presence of the absorption notch near 4660 Å, well ac-counted for by Hβ in our resulting synthetic spectra.

3.2. Type Ic SNe: the representatives

∗ SN 1987M:In Fig. 26 the observed spectrum of SN Ic 1987M, near

day 20, is compared to the generated SSp with vphot =

7000 km s−1 and Tbb = 4500 K. The OI 7773 Å trough isclearly deep and is well matched with our SSp (τ = 5).Identifying He I lines in this object is problematic and wasa subject of different investigations (Jeffery 1991; Clocchiattiet al. 1996b). In fact, Jeffery et al. (1991) have presented asynthetic-spectrum analysis around maximum and came to theconclusion that He I lines may be present. In addition they alsoclaimed the presence of a weak Hα trough. In our SSp-LTEapproach applied to the 20 day spectrum, we tested two possi-bilities for He I identification, namely He I lines at high veloc-ity (vcont(He I) = 10 000 km s−1; τ(He I) = 0.16) and a lowervelocity case (vcont(He I) = 2000 km s−1; τ(He I) = 0.6). Theupper panel in Fig. 26 corresponds to the first case, in whichthe strong P-Cygni profile at ∼5750 Å is due to Na I linealone. The high velocity He I 5876 Å then contributes to thetrough blueward of the Na I profile. The undetached Sc II linesare in part responsible for the two absorption features indi-cated in the plot (τ(ScII) = 0.7). The other He I optical lines,at 6678 Å and 7056 Å, both contribute to the strong Fe II fea-tures (τ(FeII) = 22). In the low velocity case (Fig. 26; lower


panel) a weak He I absorption at ∼6500 Å appears, whichprovides a better match to the observed spectrum, as does“He I+Na I D” at ∼5750 Å. It is not simple to decide which fitto adopt from analyzing only one spectrum. The two possibili-ties seem plausible and both indicate the presence of He I linesin SN 1987M at the age of 20 days.

We find no need to include either Si II and/or C II or Hα.We note however that Si I lines may help to fit the deep trougharound 6900 Å, but they would introduce then various un-wanted features in the rest of the spectrum.

∗ SN 1994I:SN 1994I is considered the best observed “normal” type Ic

event. Various works have presented detailed analysis of lineidentifications in this object (Wheeler et al. 1994; Filippenko1995; Clocchiatti et al. 1996b; Millard et al. 1999). The mostcontroversial issues concerned traces of H and He in the earlyspectra. Filippenko et al. (1995) have attributed the observedabsorption feature near 10 250 Å to the infrared He I 10 830 Å,arguing for the presence of helium in the ejecta of SN 1994I.Millard et al. (1999), however, clarified the related issues bymeans of synthetic spectra. One of the most important resultswas the incompatibility of the simultaneous fit of the∼10 250 Åobserved feature with He I 10 830 Å and with the opticalHe I lines. Contrary to the case of SN 1999ex (see Fig. 19),trying to account for the observed absorption near 10 250 Åwith He I 10 830 Å introduces too strong He I features in theoptical region (Millard et al. 1999). Even with a NLTE syn-thetic spectra analysis, Baron et al. (1999) were not able to re-produce the ∼10 250 Å infrared feature assuming it to be theHe I 10 830 Å line. Instead, the feature in question may beaccounted for by lines of C I and/or Si I (Baron et al. 1999;Millard et al. 1999).

Figure 27 compares the observed spectrum around maxi-mum with an SSp that has vphot = 12 000 km s−1 and Tbb =

7000 K. The match is quite good. The trough near 5700 Å isaccounted for by the undetached Na ID line (τ = 1). The otherprominent features are well described by lines of Fe II(τ = 10),Ca II(τ = 300), O I(τ = 0.7), [O II](τ = 0.2). Lines ofMg II, τ = 1, have been introduced to help the fit bluewardthe IR-Ca II P-Cygni profile. Lines of Ti II, τ = 0.8, are alsoconsidered to help the fit in the 4000−4500 Å region.

Concerning the feature near 6180 Å, it was difficult atsome epochs in the spectra presented by Millard et al. (1999),to decide between detached C II 6580 Å and undetachedSi II 6355 Å. Moreover, at −2 days and−4 days, the authors ob-tained a better fit with a combination of the two. In the spectrumdisplayed in Fig. 27 we test the two possibilities, namely unde-tached Si II (τ = 1.5) and detached C II (vcont = 8000 km s−1;τ = 1.7 × 10−3). A closer view of the region of interest is dis-played in the window. We found Si II somewhat too blue toaccount for the observed feature. There is no need at this epochto combine the two lines. Therefore we prefer detached C II asthe more probable.

∗ SN 1996aq:We analyzed two photospheric spectra of SN 1996aq,

namely one near maximum light and one at ∼24 days. The

5500 6000 6500

Fig. 27. SSp fit of SN 1994I compared to the observed spectrum atmaximum brightness. Conspicuous line features are shown. The win-dow shows the C II and Si II identification cases (see discussion).

5500 6000 6000 6500 7000

Fig. 28. SSp fit of SN 19996aq compared to the observed spectrum atmaximum light. The right and left windows show a zoomed view ofthe 6300 Å and 5700 Å regions, respectively. Conspicuous line fea-tures are also reported.

supernova has been classified as type Ic SN near maximumwhen first observed (Nakano 1996). The observed spectrumnear maximum is compared, in Fig. 28, with a synthetic spec-trum that has vphot = 9000 km s−1 and Tbb = 9000 K. TheSSp accounts for almost all the conspicuous features, namelyIR-Ca II, Fe II, O I and Mg II. The trough near 5680 Å ap-pears broader than usual. To account for this absorption fea-ture we use a combination of undetached Na ID (τ = 1) and


detached He I (vcont = 3000 km s−1; τ = 0.35). The resultingprofile fits nicely the broad trough (left window in Fig. 28). Theabsorption features due to the He I lines 6678 Å and 7065 Å aretoo weak to be clearly seen as indicated in Fig. 28. Concerningthe ∼6300 Å feature, while Si II is found to be too blue, we findthat a combination of C II (vcont = 2500 km s−1; τ = 2.2×10−3)and Hα (vcont = 6000 km s−1; τ = 0.12) provide a satisfactoryfit. On the one hand, the weak feature near 6330 Å is producedby the minimum of the C II 6580 Å absorption. This possibilityis illustrated in the right window of Fig. 28. On the other hand,a weak absorption in the emission feature near 4560 Å couldbe due to Hβ although the Hα optical depth seems too smallfor Hβ to be plausible.

Alternatively, if we try to fit the 6300 Å feature by only C II,then this latter should be expelled at very high velocity, evenmore than helium. Moreover, in this way we could not pro-duce the notch near 6330 Å. We regard the “C II+Hα” com-bination as the more probable. However a final confirmationof this is still not beyond doubt. The identification of Hα intype Ic was a subject of different discussions (Jeffery et al.1991; Filippenko 1992; Swartz 1993; Wheeler et al. 1994;Branch 2003). Generally the identification of Hα in early spec-tra of some SNe Ic objects (exp. SN 1987M and SN 1994I) hasnot been accepted. In SN 1987M for example the measured ve-locity for Hα was suspiciously low compared to calcium andoxygen. In the case of SN 1996aq we find a different situation,since the expansion velocity attributed to Hα is the highest (i.e.higher than C II and He I while the other lines are all unde-tached). It would be interesting to have more detail for thisobject and to have a larger sample of early spectra of type Icobjects.

Figure 29 compares the 24 day spectrum to the SSp thathas vphot = 4600 km s−1 and Tbb = 5000 K. Ions that are re-sponsible for the most conspicuous supernova absorption fea-tures are indicated. The match is good. Note both the strongCa II infrared triplet and Ca II H&K are well produced in theSSp. Ni II helps the fit longward of the Ca II H&K profile. Thedouble absorption feature near 9000 Å is well produced by un-detached lines of C I(τ = 1) and Mg II(τ = 3). Lines of O I and[O II](at 7321 Å) are also introduced to account for featuresnear 7640 Å and 7220 Å, respectively. A noticeable featureof the SN at this phase is the clear appearance of He I lines,that question the classification of SN 1996aq as type Ic ob-ject. The broad P-Cygni profile with the absorption minimumnear 5720 Å cannot be accounted for by Na ID alone. Wefavour He I 5876 Å because it implies other observed features,especially the He I lines at 6678 Å and 7065 Å. He I lines in theSSp are produced adopting vcont = 2400 km s−1 and τ = 0.25.

The slope in the 5950−6450 Å wavelength range doesnot contain conspicuous features although it has some weakabsorptions, reminiscent of what is seen for example inSNe 1984L and 1998T at similar phases. We regard SN 1996aqas a transition object between Ib and Ic, rather than a “pure”type Ic event. We have checked different line combinations inorder to decide which ions to introduce to fit the 5950−6450 Åregion, and especially the feature near 6355 Å. The differentpanels in Fig. 29 illustrate these possibilities. In fact, whilelines of Fe II and Si II(undetached) provide a good fit to the

Fig. 29. SSp fit of SN 1996aq compared to the observed spectrumaround day 24. The different panels show the possible line identifi-cations for the 6300 Å absorption, namely C II (upper panel), Hα(middle panel) and Ca I (bottom panel). Conspicuous line features arereported.

observed features, we find it difficult to decide between C II,Hα and Ca I to account for the absorption near 6355 Å.The three cases correspond respectively to: C II with vcont =

5400 km s−1 and τ = 7 × 10−6 (upper panel); Hα withvcont = 4900 km s−1 and τ = 0.02 (middle panel); undetachedCa I with τ = 6 (lower panel). The Ca I case, lower panel,produces an unwanted absorption trough near 6060 Å. Forthe C II case, upper panel, the element is assigned a veloc-ity 3000 km s−1 greater than the one of helium, which is hardto accept. Therefore the most probable situation remains thecase of the high velocity hydrogen (middle panel).

3.3. Type IIb SNe: the representative

∗ SN 1993J:The discovery of the “hybrid” type IIb SN 1993J has cre-

ated a link between SN Ib-c and type II objects, creating andextending our understanding of the physics of core-collapse. Atearly phases, this SN displays conspicuous hydrogen Balmerfeatures similar to type II SNe. At the nebular phase how-ever, the spectrum shows many signatures of type Ib-c SNe,namely strong oxygen and calcium lines (Filippenko 1994;Lewis 1994; Matheson et al. 2000). Furthermore, the analy-sis of late epoch spectra reveals that traces of hydrogen (i.e.Hα in emission were still present (Patat et al. 1995). Anotherwell observed SN belonging to type IIb class is SN 1996cb. Bycomparison with SN 1993J, SN 1996cb showed Balmer lineswith stronger P-Cygni profiles. In addition, the photosphereof SN 1996cb receded faster than its counterpart in SN 1993J


Fig. 30. SSp fit of SN 1993J compared to the observed spectra at16 days (upper panel), at 24 days (middle panel) and at 59 days (lowerpanel). Conspicuous line features are reported.

and the He I features (in absorption) appeared earlier, aroundday 24 in SN 1993J and near maximum light in SN 1996cb(Qiu et al. 1999; Deng et al. 2001).

We have analyzed spectra at three different epochs forSN 1993J, namely at 16, 24 and 59 days. Figure 30 displaysour best fit spectra compared to the observe ones. The plottedsynthetic spectra have vphot = 9000 km s−1 and Tbb = 7800 K(16 days; upper panel), vphot = 8000 km s−1 and Tbb = 7000 K(24 days; middle panel) and vphot = 6000 km s−1 and Tbb =

5000 K (59 days; lower panel). The fit to the 16 day spectrumis good, except the Hα profile. The corresponding strong andbroad P-Cygni feature cannot be produced completely by the“SYNOW” code. This is because in our SSp treatment we areadopting a resonance scattering source function. The hydro-gen reference line has vcont = 1000 km s−1, while an opticaldepth of τ = 21 is used. Profiles of Hβ, Hγ and Hδ are clearlydiscernible. The P-Cygni profile around 5700 Å is assigned toHe I 5876 Å, with a small contribution from Na ID. The otheroptical He I lines are not visible at this phase. The helium isfound to be undetached and has a moderate optical depth of 0.6.The shallow IR-Ca II triplet is fitted adopting τ = 20. The ob-served Ca II H&K is nicely reproduced in the SSp.

As the supernova evolves, the IR-Ca II profile becomesclearer (Fig. 30; middle panel). At this phase an optical depth ofτ(CaII) = 300 is used. Fe II lines are also clearer compared tothe 16 day spectrum (τ(FeII) = 3 at 16 days and τ(FeII) = 12at 24 days). The notch appearing on the emission componentof Hα is attributed to He I 6678 Å. Helium is still undetachedwith an optical depth of 1.5. SN 1996cb at a similar phase, near25 days, has more evident He I lines compared to SN 1993J,

especially the lines at 6678 and 7056 (Fig. 2; Deng et al.2001). In fact, its 25 day spectrum resembles the 59 day spec-trum of SN 1993J rather than the 24 days one. The He I lines atday 59 are prominent (Fig. 30; bottom panel). The He I refer-ence line is undetached and has τ(He I) = 30. The correspond-ing SSp fit is quite good for the 5876 Å line, while it is difficultto obtain a good match with the other optical He I lines.

Hydrogen Balmer lines become weak at this phase.Ba II lines are introduced to improve the fit just blueward ofHe I 5876 Å (τ(BaII) = 5). At this phase, forbidden emis-sion lines develop, indicating the transition to the nebular phase(exp. [O I] 5577 Å, [O I] 6300,64 Å and [Ca II] 7291,7324 Å;bottom panel). The O I 7773 Å absorption profile is repro-duced at the three different epochs by τ = 0.3(at 16 days),τ = 0.4(at 24 days) and τ = 2.5(59 days). Ni II lines are also in-troduced in the 24 days and 59 days synthetic spectra, blendedwith Hδ, to improve the match with the observed weak absorp-tion on the Ca II H&K emission component.

3.4. Type II SNe: the representative

∗ SN 1999em:

As a representative of the type II class, we have analyzedSN 1999em. The well sampled observations, spectra and pho-tometry, have been presented and studied by Elmhamdi et al.(2003). The event shows characteristics typical of a type II-Pobject, namely clear and broad Balmer P-Cygni profiles and aplateau phase of almost constant luminosity in the optical.

Figure 31 displays two spectra during the photosphericphase. For comparison with other events of our CCSNe sam-ple, the phases are normalized to maximum light in type Ib-cSNe, adopting 16 days as the rise time to reach maximum. Theobserved spectra are compared to the best fit synthetic ones thathave vphot = 10 000 km s−1 and Tbb = 10 000 K (−6 days; up-per panel) and vphot = 4600 km s−1 and Tbb = 6400 K (25 days;lower panel). Synthetic line features are designated accordingto the ion whose line gives rise to the feature (Fig. 31). Atthe earliest phase, i.e. −6 days, only three elements are intro-duced in the SSp model. Indeed undetached lines of Balmerhydrogen, He I and a weak contribution from Ca II are almostsufficient to reproduce the most conspicuous features super-imposed on the “hot” continuum. The fit with Na I D, at thisphase, is poor compared to He I 5876 Å. Moreover He I con-tribution, in type II SNe, is found to be important shortly afterthe explosion. In fact in SN 1987A, the He I 5876 Å featurewas clearly present during the first few days, then owing tothe decreasing envelope-temperature it rapidly faded and dis-appeared completely around 1 week after the explosion, whenNa I D starts to emerge (Hanuschik 1988). Baron et al. (2000),in analyzing very early spectra of SN 1999em, have found ev-idence for helium enhanced by at least a factor of 2 over thesolar value. The interpretation of the He I strength in termsof helium-overabundance is however premature, since freeze-out effects were not considered and could lead to the enhancedhelium-excitation compared to the steady-state model (Utrobin2002).


Fig. 31. SSp fit of SN 1999em compared to the observed spectra atday −6 (upper panel) and at day 25 (lower panel). Conspicuous linefeatures are shown. For comparison, the reported phases are normal-ized to maximum in type Ib-c SNe, using “16 days” as the rise time.

Balmer hydrogen features, with τ = 15, are clearly evident.As in the case of SN 1993J, we cannot reproduce the full ob-served Hα profile for the same reason. At day 25, Hα P-Cygniprofile becomes narrower because of the decrease in the ex-pansion velocity (Fig. 31; bottom panel). Hα in the SSp hasnow vcont = 1000 km s−1 and τ = 21. The envelope tempera-ture decreases and many lines emerge at this phase. Apart fromthe hydrogen lines (slightly detached), all the lines are unde-tached. The match with almost all observed features is good.The Ca II H&K and IR-Ca II are nicely reproduced. A few dis-tinct absorption lines redward of the near infrared Ca II tripletare identified as C I, τ = 0.6, through our spectral synthesis.Sc II lines, τ = 3, account rather well for features near 5450,5600 and 6170 Å. The blue part of the spectrum, λ < 5000 Å,is subject to severe line blending, but it seems well matchedwith contributions from lines of Fe II(τ = 10), Ni II(τ = 5),Ti II(τ = 8), Ba II(τ = 2), Mg II(τ = 2), Ca II and hydrogen.The feature near 7670 Å is attributed to O I 7773 Å(τ = 0.3),while there may be a contribution from [O II] around 7230 Å.

4. Spectroscopic mass estimates

In the following we adopt two methods in order to obtainrough estimates of some ejecta masses (i.e. the total ejecta massand/or a given element mass). The methods use the results fromour spectral best-fits.

4.1. Method 1

We compare the optical depths derived from our best-fits to theoptical depth plots for different compositions presented in the

work by Hatano et al. (1999), making sure of course that thesame reference lines are used. In that paper the authors pre-sented a systematic survey of ions that could be responsible forsupernova features in six different compositions. The LTE op-tical depth of each reference line, of a given ion, is computedand then plotted against temperature. We focus on the ratio ofhydrogen to helium optical depths, i.e. τ(H I)/τ(He I), attempt-ing thereafter to translate this ratio into a relative abundance ofthe two elements. This is important as it can give us a roughidea about the amount of hydrogen present in these objects.The optical depth ratio is assumed to trace the abundance ra-tio of the two elements. This is essentially due to the fact thatthe τ-calculations of Hatano et al. (1999) adopted a fixed elec-tron density. The τ(H I)/τ(He I) ratio, indeed, depends on elec-tron density, temperature and hydrogen to helium abundanceratio. Therefore for a given temperature, the τ(H I)/τ(He I) ra-tio depends in principle only on the abundance ratio. We mustremember however that the “τ-plots” are for LTE, with no al-lowance for non-thermal excitation is made. For helium, forexample, non-thermal excitation effects are significant in de-veloping He I features (Lucy 1991). More detailed estimatesmight take this effect into account, if one has a clearer under-standing of outward mixing of radioactive material.

For SN 1990I around maximum and according to our bestSSp fit, the optical depth ratio, evaluated at the same veloc-ity3, is measured to be τ(H I)/τ(He I) ∼ 0.6. At similarhigh temperatures (i.e. ∼14 000 K), one obtains a value ofabout 40 for that ratio according to the optical depth plots ofdifferent ion reference lines for a hydrogen-rich compositionin Hatano et al. (1999) (see their Fig. 2). In addition, in thecited work the τ(H I)/τ(He I) ratio remains almost constant athigh temperatures, then it starts increasing once the tempera-ture decreases. For example around T ∼ 10 000 K, the ratio isabout 80. Near 9000 K, however the ratio reaches a value ofabout 1000. These values (i.e. 40, 80 and 1000 at their corre-sponding temperatures) correspond to a factor of 10.23 in abun-dances between hydrogen and helium (Table 1; Hatano et al.1999). Assuming proportionality, we obtain then a rough esti-mate of the hydrogen to helium relative-abundance for the caseof SN 1990I around maximum. In fact a relative abundanceof ∼0.15 is found. For SN 2000H in which Hα is clearly deep(τ ∼ 5 at maximum; Fig. 23), an abundance ratio (i.e. H/He)is estimated to be ∼0.23. SN IIb 1993J, around day 16, has aratio ∼0.8. We estimate an even higher ratio for SN 1999em,namely ∼3.2. These estimates, although rough, are in accordwith our impression that the hydrogen mass increases as we gofrom type Ib to IIb to type II SNe. Figure 32 displays the log-arithmic hydrogen to helium abundance ratios against temper-ature for the events of our sample. Note here that for low tem-peratures, T ≤ 9000, the reported values in Fig. 32 can be takenas upper limits. This is because we used a value of 1000 for the

3 The optical depths reported in Table 2 are evaluated at the detach-ment velocity of the corresponding reference lines (e.g. for SN 1990Iat maximum τ(H I) at 16 000 km s−1and τ(He I) at 14 000 km s−1). Fora “v−8” power-law, the optical depth of He I evaluated at the same ve-locity of H I is then:τ(HeI) = 2.9 × (14/16)8 � 1.0.


Fig. 32. The hydrogen to helium abundances ratio against temperature.See Sect. 4 for more details.

ratio τ(H I)/τ(He I). At low temperatures, however, the ratiobecomes significantly larger than this adopted value (Fig. 2a;Hatano et al. 1999). An additional source of uncertainty maybe related to the continuum temperature estimates owing to thetotal reddening effects. Furthermore, according to our fit ex-perience on the sample spectra, an uncertainty of about 10%is assigned to the derived values of the optical depths. In thecase of SN IIP 1999em and SN IIb 1993J, with their com-plete Hα P-Cygni profiles, greater optical depth uncertaintiesshould be expected, although in these events we got good fitsto the Hβ profiles. It is worth noting however that the enormousNLTE effects in the He I lines lead to very inaccurate determi-nation of helium abundances.

Making use of the derived photospheric velocities “vphot”,from our synthetic spectra modeling, it is possible to recover“spectroscopic” estimates of the kinetic energy and mass abovethe photosphere. In fact, Millard et al. (1999) have shown thatfor spherical symmetry and an“r−n” density distribution, themass (in M�) and energy (in 1051 erg) above the electron-scattering optical depth “ τes” can be expressed as:

M = (1.2 × 10−4) v24 t2d µe τes fM(n) (1)

E = (1.2 × 10−4) v44 t2d µe τes fE(n); (2)

where fM(n) = n−1n−3 and fE(n) = n−1

n−5 , td is time after explosionin days, v4 is vphot in units of 104 km s−1 and µe is the meanmolecular weight per free electron.

In applying the above equations to our supernova sam-ple, we adopt n = 8 and τes = 2/3. For the mean molecularweight per free electron we have different values for each case,namely: SNe Ib: µe = 8 for half ionized helium or doubly ion-ized oxygen; SNe Ic: µe = 14 for a mixture of carbon and oxy-gen, both singly ionized; SNe II: µe = 1 to 2 for fully ionizedor half ionized hydrogen. For the case of “hybrid” IIb SNe weassume a mixture of hydrogen and helium, both half ionizedand hence µe = 5.

Fig. 33. The evolution in time of the “spectroscopic” estimates ofMass (upper panel) and Energy (lower panel) above the photosphere.

Figure 33 displays results for five events. Filled symbolsrefer to the mass (in units of M�; upper panel), while the openones indicate the kinetic energy (in units of 1051 erg; lowerpanel). For each supernova, the estimated masses for differentepochs are connected by a dotted line. The short-dashed lineconnects the derived energies in order to clarify trends. For agiven phase, the reported amounts indicate the mass movingabove the photospheric velocity, carrying a corresponding ki-netic energy. The events displayed in Fig. 33 have been selectedon the basis of their type and velocity. SN 1990I, being an ob-ject with high velocities, appears to have high kinetic energy,while SN 1991D with its lower velocity structure lies at the bot-tom of the plot. For a significant comparison, we choose data atsimilar phases, around day 38. At that epoch, the mass movingfaster than “vphot” and the corresponding energy are estimatedto be: (1.1 M�; 1.6 f oe) for SN 1990I; (0.45 M�; 0.28 f oe) forSN 2000H; (0.3 M�; 0.1 f oe) for SN 1991D; (0.6 M�; 0.8 f oe)for SN 1993J and (0.95 M�; 0.77 f oe) for SN 1994I.SN Ib 1984L has behaviour similar to SN 2000H. Thesederived quantities suffer, of course, from some uncertain-ties, namely those incurred with the derived photosphericvelocities4 from our best fits and the adopted value for thepower-index “n”. However, they seem to give reasonable andmeaningful values that are not in disagreement with othermethods (e.g. light curve modeling; NLTE treatment of earlyand late spectra), especially for the well studied objectssuch as SNe 1990I, 1994I and 1993J. The oxygen-mass es-timates could be a further check of the trend in the derivedphotospheric outflow mass. Elmhamdi et al. (2004), on the

4 A mean fitting uncertainty of ∼500 km s−1 is assigned to the de-rived velocities. The uncertainty in the velocity would affect more theuncertainties in the derived energies than those of the derived masses.


basis of [O I] 6300, 6364 Å line analysis at nebular phases,have estimated a lower limit on the oxygen mass to fall inthe range 0.7−1.35 M�. The available estimated amounts forthe other objects are as follows: ∼0.3 M� for SN 1984L(Filippenko 1990); ∼0.5 M� for SN 1993J (Houck & Fransson1996) and ∼0.4 M� for SN 1994I (Woosley et al. 1995).These estimates follow the trend seen in the masses above thephotosphere.

We emphasize here that this simple approach, usingEqs. (1) and (2), does not provide “reasonable” estimates in thecase of type II SNe. In fact when this method is used for bothSNe 1987A and 1999em, the estimated “M” and “E” abovethe photosphere seem to be very small, in contradiction withwhat we may expect. Two possible effects might invalidate themethod for type II objects. First, because the method cannot beused whenever the power index “n” value falls below 5 (i.e. tokeep fM(n) and fE(n) positive and convergent). For SN 1987A,indeed, Jeffery & Branch (1990) had found “n” falling below 5already within a week after the explosion (see their Figs. 15and 19). Second, in type II SNe and after the earliest times,the electron-scattering optical depth for the photosphere comesfrom a thin layer of ionized hydrogen at the recombinationfront. Farther out, there can be a lot of neutral hydrogen (andhelium) that is not accounted for in the “Millard et al.” equa-tions adopted in our simple analysis. When, for instance, “µe”is set to one, the equations are only giving the mass and en-ergy of the matter near the photosphere. Consequently onemay question why they should be taken seriously for the otherCCSNe classes. Some carbon and oxygen could be neutral too.One simple explanation could be that because of the large hy-drogen envelope, these effects seem to be more “severe” intype II events rather than the rest of SNe. In summary, the sim-ple method does not apply to SNe II because it does not takeinto account the recombined hydrogen.

By adopting the estimate of the outflow mass above “vphot”as an upper limit of the helium mass in type Ib and IIb events,and using the derived constraints on the hydrogen to heliumabundances ratio, one can thus recover an upper and roughestimate of the mass of expelled hydrogen. Around maxi-mum light, the hydrogen amount in SN 1990I is found tobe ∼0.16 M�. At similar phase, an amount of about 0.1 M�is computed for SN 2000H. SN 1993J, however, seems to ejecta quantity as high as ∼0.7 M�. SN 1993J indeed serves as acontrol case for our adopted methodology. Different studiesargued that a low-mass hydrogen envelope on a helium coreis the most likely scenario for the progenitor of SN 1993J.On the one hand, light curve modeling demonstrates a com-patibility fit with a 4 M� helium core and hydrogen envelopeof ∼0.2 M� (Woosley et al. 1994). A similar analysis invoked ahelium core ∼4−6 M�, with a residual hydrogen mass less than0.9 M� (Nomoto 1993; Shigeyama et al. 1994). On the otherhand, best fit models of spectra lead to similar progenitor prop-erties, namely 3.2 M� for the helium core with a 0.2−0.4 M�hydrogen envelope (Houck & Fransson 1996). Our estimate istherefore in good agreement with the previous cited studies.For the rest of type Ib objects, an upper limit of the hydrogenmass of the order 0.1 M� is estimated.

4.2. Method 2

Alternatively, an approximate way to estimate the hydrogenmass invokes the amount needed to fill a uniform densitysphere of radius “v×t” at an epoch td since explosion. Adoptinga value of “10−9” for the hydrogen Balmer fraction (within theLTE context), the required ion mass can be given by:

M(M�) � (2.38 × 10−5) v34 t2d τ(Hα); (3)

This comes from the equation for the Sobolev optical depth foran expanding envelope (see e.g. Castor 1970; Jeffery & Branch1990). where td in days and v4 is in 104 km s−1. This equation isthen used to determine the amount filling a spherical shell of aninner radius “vmin × td” and an outer edge of radius “vmax × td”.Constraints on the shell widths, “vmin− vmax”, are approximatedusing Gaussian fits to the observed Hα absorption troughs bymeans of the corresponding FWHM. It is worth noting here thatconstraints on “vmax” from this method, i.e. Gaussian fit, can beconsidered as a lower limit to the “real” maximum velocity.

Although non-thermal excitation and “NLTE” effects maybe also important for hydrogen, the method seems to give rea-sonable estimates. For SN 1990I around maximum light a valueof 0.02 M� is computed, while at similar phase SNe 1983Nand 2000H have, respectively, 0.008 and 0.08 M�. For type IIbSN 1993J, at day 16 since maximum, the method gives anapproximate amount as high as 1.7 M�. This amount is ofcourse very sensitive to the assumed optical depth. In factSN 1993J has an optical depth similar to type IIP 1999em (i.e.τ(Hα) = 21 at ∼25 days). However the fit is more “convinc-ing” in SN 1999em than in SN 1993J, especially in the Hα ab-sorption troughs (see Fig. 30-middle panel and Fig. 31-lowerpanel). This might be indicative of an overestimated opticaldepth in SN 1993J and/or a larger departure from “LTE” com-pared to SN 1999em.

A rise time of 15−20 days has been adopted in this classof object. Moreover, if we adopt a representative Hα opticaldepth of 0.5 at day 20 with an Hα velocity restriction sim-ilar to SN 1990I, the estimated hydrogen mass is of the or-der 0.015 M�. Events with higher velocity widths and/or deeperHα troughs would eject larger amounts.

5. Discussion and conclusion

One of our main goals was to identify traces of hydrogen intype Ib-c, and to identify any systematic similarities and differ-ences correlating with other physical properties. Consequently,guided by modeling early spectra of SN Ib 1990I, we haveexplored different combinations of ions and shaping the6000−6500 Å wavelength range. Special attention has been de-voted to the feature seen near 6300 Å. Even though the numberof events with spectra well sampled in wavelength, is limited,it appears that hydrogen in varying amounts is identifiable inmost type Ib objects especially at very early times. Only in twocases, namely SNe 1991D and 1991L, does the Ne I 6402 Åline remain as an alternative possibility with large departurefrom LTE. Even in SN 1996aq, that we classified as a transitiontype Ib/c event, and SN 1999ex the presence of the Hα troughseems highly preferred.


Fig. 34. Comparison of SN 1990I with SNe 1984L and 2000H, at twodifferent phases of evolution. The vertical dashed lines indicate thelocations of Hα and Hβ troughs in SN 2000H. He I optical series aremarked by vertical ticks.

At later phases, more than ∼3 weeks, SNe 1984L,1988L, 1991ar, 1998dt, 1998T and 1999dn behave quite sim-ilarly, namely still showing evident He I lines and a “flat”6000−6500 Å region of the spectrum, with a weak Hα ab-sorption feature. Even SN 1991D with its particularly lowphotospheric velocity and narrow lines belongs to this class.SN 1991L appears to be similar to SN 1991D in having shal-low He I optical features. SNe 1999di and 2000H are the uniqueevents among the sample that display obvious He I lines, to-gether with a pronounced and deep Hα absorption line. It is in-teresting to note here the peculiar behaviour of SN 1997dc. Theevent has He I troughs as deep as in SNe 1999di and 2000H,however it has a flat 6000−6500 Å region. SN 1990I at thisphase shows distinct He I lines and a moderate Hα trough.Figure 34 compares SN 1990I with SNe 1984L and 2000H attwo different phases of evolution. The vertical dashed lines in-dicate the locations of Hα and Hβ troughs in SN 2000H. Onthe one hand, while at early phases (upper panel) the troughassigned to Hα is clear in all the three SNe, it disappears lateron in SN 1984L (lower panel). On the other hand, Hβ appearsclearly in SN 2000H and absent in SNe 1984L and 1990I.Moreover, the comparison emphasizes the high velocity be-haviour of SN 1990I as is evident from the blueshifted fea-tures in SN 1990I compared to the others. He I optical lines arealso shown. For the three events the He I troughs at 5876 Å,and 7065 Å are clearly identified and get narrower with time.At early phase the He I 6678 Å is recognized in SNe 1990Iand 1984L while it is just hinted in SN 2000H. Later on,Fig. 34-bottom panel, the He I 6678 Å line grows in strength inSNe 1984L and 2000H while it fades in SN 1990I. The compar-ison of these three events, in terms of changes in line visibility

and velocity, demonstrates the complexity involved in demon-strating any thing like a continuous sequence.

Hydrogen manifests its presence in a different way in theother CCSNe types. A part from SN 1996aq that we re-classifyas a transient Ib/c event rather than a pure type Ic SN, thereis no evidence for the presence of hydrogen in type Ic ob-jects analyzed here (i.e. SNe 1987M and 1994I). The “hybrid”SN IIb 1993J displays typical type II features at early phases,such as strong and broad Hα P-Cygni emission component,which is absent in type Ib objects. This can be explained withinthe context of the “detachment” concept. In fact, Hα P-Cygniprofile would lose its obvious emission component when itis highly detached. At early epochs, both SN IIb 1993J andSN II 1999em show clear evidence of the other H I Balmer linessince the corresponding optical depths are too large to allowthem to be distinctly visible, contrary to normal type Ib objects,with the exception of some cases with deep and conspicuousHα troughs that present signature of Hβ too (e.g. SNe 1999diand 2000H). Two factors make Hβ barely discernible in type Ib:the optical depth found to fit Hα is small and the contrast veloc-ity of Hα is high. The opposite is seen in type II and IIb events.As time goes on, the Hα emission peak in SN 1993J changesto a double-peaked structure which we recognized as the emer-gence of the He I 6678 Å, with more clearer He I optical lines.These laters disappear at the nebular phase and the spectrum isdominated by typical Ib features such as [O I], [Ca II] and Ca IIin emission. The situation in SN 1999em, and in type II in gen-eral, is different. The He I lines are found to disappear veryearly, about one weak after explosion, and even at later neb-ular epochs the Hα is still the most prominent feature in thespectrum.

In Fig. 35, upper panel, we report the resulting photo-spheric velocities from our best fits for the CCSNe sample.Data for SN 1987A, corresponding to Fe II 5018 Å absorp-tion, are also shown for comparison (dashed line; Phillips et al.1988). Additional points for SN 1994I are taken from Millardet al. (1999). The plot indicates the low velocity behaviour oftype II SNe, both 1987A and 1999em, at early as well as in-termediate epochs. SN IIb 1993J follows somewhat similar be-haviour as SN Ic 1994I in having higher velocities. As far astype Ib SNe are concerned, they appear to display a differentvelocity evolution. The scatter seems to increase at intermedi-ate phases (around 20−30 days). This fact can be simply due tothe paucity of available observations outside that range. Aroundday 20, for example, a scatter as high as 5000 km s−1 is mea-sured. SNe 1990I and 1998dt belong to a class with the higher“vphot”, while objects such as SNe 1991D and 1996aq have thelowest estimated velocities, approaching even type II objects.The remainder of the Ib events follow a similar trend, namelythe one described by Branch et al. (2002). SN 1999ex is foundto belong to this class. Surely more data are needed to obtainmeaningful statitical conclusions.

The middle and bottom panels in Fig. 35 display the evo-lution of the Hα contrast velocities, “vratio

cont ” and “vcont” respec-tively, for the sample. As discussed before, when discussingdifferences in the Hα profile in CCSNe, we would expectan increasing value of “vcont(Hα)” going from type II to IIbto Ib SNe. This trend is in fact illustrated in the plot (bottom


SN87A(II)SN93J(IIb)SN94I(Ic)SN99em(IIP)SNeIb

SN99emSN93JSNeIb

Fig. 35. The photospheric velocity evolution of Ib SNe sample, com-pared to SNe 1987A, IIP 1999em and Ic 1994I (upper panel). Middlepanel: the Hα contrast velocity ratio evolution (i.e. vratio

min (Hα)/vphot).Lower panel: the Hα contrast velocity evolution (i.e. vmin(Hα)− vphot).Results for SN IIP 1999em and SN IIb 1993J are also shown.

panel). A similar trend is seen in the “vratiocont (Hα)” evolution

as indicated by the middle panel. While in SN IIb 1993Jand SN IIP 1999em the line is found to be either unde-tached or slightly detached, it is highly detached in type Ibevents. Moreover, the “vcont(Hα)” is found to increase withinthe first 15 days, reaching values as high as 8000 km s−1, andthen follows an almost constant evolution. According to Table 2and up to ∼60 days, type Ib SNe have hydrogen down to11 000−12 000 km s−1, while in SN 1993J hydrogen is downto 8000 km s−1. SNe II appear to have hydrogen down to evenlower velocities (∼5000 km s−1 in SN 1999em). In addition, hy-drogen in type Ib SNe is found to have very small optical depthsindependently of the contrast velocity (also independently ofthe phase). This is shown by Fig. 36. In fact, type Ib objectspopulate the region of extremely low optical depths, althoughthe “vratio

cont (Hα)” spans a large range, contrary to SNe 1993Jand 1999em.

Table 2 also displays properties of the helium line. Intype Ib SNe, He I lines are found to be not always detachedas is the case for hydrogen. The He I 5876 Å line is found to beclearly distinguishable in type Ib objects. A contribution fromNa ID is argued for in some cases, such as the broad featurein SN 1996aq and cases with an observed bump shortward ofthe minimum absorption as seen in SNe 1990B and 1997dc.It is important to recall here that in some cases we obtaineda better fit to the He I optical lines adopting an e-folding SSp,which allows a two-component treatment of the line and a grad-ually decreasing optical depth below the detachment velocity,

Fig. 36. The Hα velocity contrast evolution, vratiocont (Hα), versus Hα op-

tical depth. Results for SNe IIP 1999em and IIb 1993J are displayedfor comparison. Note that results are displayed independently of thephase.

rather than a discontinuity in it. In the case of SN 1999ex,we extend our He I line fits beyond 1 micron, confirming thepresence of various IR lines, namely at 1.083, 1.284, 1.700(?),and 2.085 microns (Fig. 20), in accord with optical He I pa-rameters. In our analysis, we also pointed to some interestingline identifications. For example the contribution of Sc II linesin nearly all the CCSNe spectra, especially for reproducing thedouble absorption features blueward of the He I 5876 Å line.We note here that except for He I and H I, almost all the linesin the CCSNe synthetic spectra are undetached.

Another important study is the behaviour of the O I 7773 Åline. Figure 37 reports the resulting optical depths correspond-ing to our best synthetic spectra fits. It is not easy to draw clear-cut conclusions from this figure since one needs, for instance,to populate the figure with more type Ic objects. However, atintermediate phases, it seems that type Ib objects tend to con-centrate in the low optical depth region, while SN Ic 1987M isfound to display the deepest profile. SNe 1993J and 1999em,at similar phases, are the objects with the lowest O I 7773 Åoptical depth. At somewhat later epochs, transient type Ib/c ob-jects display deep O I 7773 Å troughs. The stronger and deeperpermitted oxygen lines at early phases of SNe Ic and Ib/c spec-tra might imply that they are less diluted by the presence ofa helium envelope. Indeed one might expect oxygen lines tobe more prominent for a “naked” C/O progenitor core. Despitethe paucity of well sampled CCSNe observations, two obser-vational aspects tend to reinforce this belief: First, the forbid-den lines, especially [OI]6300, 6364 Å, seem to appear earlierfollowing a SNe sequence “Ic−Ib−IIb−II”. In fact the oxygenline emerges at an age of 1−2 months in type Ic SN 1987M(Filippenko 1997). SN Ic 1994I displayed evidence for the lineat an age of 50 days, although some hints may even be seen in


Fig. 37. The optical depth of the O I 7773 Å line according to the bestfit spectra. Data for different SN types are reported for comparison.

the ∼36 day spectrum (Clocchiatti et al. 1996b). While in SN Ib1990I it was hinted at the 70 day spectrum (Elmhamdi et al.2004). In other type Ib SNe it appears earlier than in SN 1990I.In SN IIb 1993J, a transition object, the line was visible in the62 day spectrum (Barbon et al. 1995). SN 1996cb, another wellobserved IIb event, showed evidence of the [OI]6300, 6364 Åline around day 80 (Qiu et al. 1999). In SNe II, however, theline appears later: around day 150 in SN 1987A (Catchpole1988) and after day 138 in SN 1992H (Clocchiatti et al. 1996a).In SN II 1999em it is suggested at a somewhat earlier phasecompared to SNe 1987A and 1992H, namely at day 114.Whether all this can be understood in terms of the lower pro-genitor mass of SN 1999em and its presumed lower oxy-gen mass remains to be investigated (Elmhamdi et al. 2003).Second, it seems that the nebular emission line decreases inbreadth following the SNe sequence above. Of course muchwork is still needed in this respect based on larger CCSNe sam-ples at both photospheric and nebular phases, relating them tothe photometry of the CCSNe variety (Elmhamdi & Danziger,in preparation). Furthermore, one important point is to checkfor any possible correlations between the oxygen minimum ve-locity (at early epochs), their line widths, and to relate thisto a progenitor mass indicator [Ca II]/[O I] ratio (at nebularphases; Fransson & Chevalier 1989), the progenitor properties(i.e. masses, energies). Other factors such as mixing, variationin envelope densities and metallicity may play an additionalcomplicating role.

Finally we presented two methods to determine the ejectaand hydrogen mass in CCSNe and especially in type Ib events.Although the methods are very approximate5, our results donot conflict with more detailed estimates, those based on

5 “Method 2” is a direct method, while “Method 1” is undirect andgives upper limits on the hydrogen mass.

hydrodynamical and NLTE models. A thin layer of hydrogen,ejected at high velocities down to 11 000−12 000 km s−1, ap-pears to be present in almost all the type Ib events studied here.These results suggest possible directions for further more so-phisticated work. The necessity to introduce lines of Ne I, Sc II,Ba II, Ca I, Fe II in some cases but not in others must inevitablyraise concerns about the identifications and other reasons forthese variations.

Acknowledgements. A. Elmhamdi is grateful to the ESO (EuropeanSouthern Observatory) support that allowed his stay at Garchingwhere part of the work has been done. I. J. Danziger was supportedin this work by MIUR with a grant for PRIN 2004 from the ItalianMinistry of Education. Research on supernovae at Harvard Universityis supported by NSF Grant AST-0205808. Supernova research atOklahoma University is supported by NSF grants AST-0204771 andAST-0506028, and NASA LTSA grant NNG04GD36G.

References

Barbon, R., Benetti, S., Cappellaro, E., et al. 1995, A&AS, 110, 513Baron, E., Branch, D., Hauschildt, P. H., Filippenko, A. V., &

Kirshner, R. P. 1999, ApJ, 527, 739Baron, E., Branch, D., Hauschildt, P. H., et al. 2000, ApJ, 545, 444Benetti, S., Branch, D., Turatto, M., et al. 2002, MNRAS, 336, 91Branch, D. 1972, A&A, 16, 247Branch, D. 2001, in Supernovae and GRBs: The Largest Explosions

in the Univers, ed. M. Livio (Cambridge Univ. Press), 96Branch, D. 2003, in A Massive Star Odyssey: From Main Sequence

to Supernova, ed. K. van der Hucht, A. Herraro, & C. Esteban(San Francisco: ASP), 346

Branch, D., Benetti, S., Kasen, D., et al. 2002, ApJ, 566, 1005Castor, J. I. 1970, MNRAS, 149, 111Catchpole, R. M., Whitelock, P. A., Feast, M. W., et al. 1988,

MNRAS, 231, 75Clocchiatti, A., & Wheeler, J. C. 1997, ApJ, 491, 375Clocchiatti, A., Benetti, S., Wheeler, J. C., et al. 1996a, AJ, 111, 1286Clocchiatti, A., Wheeler, J. C., Brotherton, M. S., et al. 1996b, ApJ,

462, 462Clocchiatti, A., Suntzeff, N. B., Phillips, M. M., et al. 2001, ApJ, 553,

886Deng, J. S., Qiu, Y. L., Hu, J. Y., Hatano, K., & Branch, D. 2000, ApJ,

540, 452Deng, J. S., Qiu, Y. L., & Hu, J. Y. 2001

[arXiv:astro-ph/0106404]Elmhamdi, A., Danziger, I. J., Chugai, N. N., et al. 2003, MNRAS,

338, 939Elmhamdi, A., Danziger, I. J., Cappellaro, E., et al. 2004, A&A, 426,

963Filippenko, A. V. 1988, AJ, 96, 1941Filippenko, A. V. 1992, ApJ, 384, L37Filippenko, A. V. 1997, ARA&A, 35, 309Filippenko, A. V., Porter, A. C., & Sargent, W. 1990, AJ, 100, 1575Filippenko, A. V., Matheson, T., & Barth, A. J. 1994, AJ, 108, 2220Filippenko, A. V., Barth, A. J., Matheson, T., et al. 1995, ApJ, 450,

L11Fisher, A. 2000, Ph.D Thesis, Univ. OklahomaFransson, C., & Chevalier, R. A. 1989, ApJ, 343, 323Gal-Yam, A., Fox, D. B., Kulkarni, S. R., et al. 2005, AJ, 630, 29Gomez, G., & Lopez, R. 2002, AJ, 123, 328Hamuy, M., Maza, J., Pinto, P. A., et al. 2002, AJ, 124, 417Hanuschik, R. W., & Dachs, J. 1988, A&A, 205, 135


Harkness, R. P., Wheeler, J. C., Margon, B., et al. 1987, ApJ, 317, 355Hatano, K., Branch, D., Fisher, A., Millard, J., & Baron, E. 1999,

ApJS, 121, 233Houck, J. C., & Fransson, C. 1996, ApJ, 456, 811Jeffery, D. J., & Branch, D. 1990, in Supernovae, Sixth Jerusalem

Winte School for Theoretical Physics, ed. J. C. Wheeler, T. Piran,& S. Weinberg (Singapore: World Scientific), 149

Jeffery, D. J., Branch, D., Filippenko, A. V., & Nomoto, K. 1991, ApJ,377, 89

Kidger, M. 1988, IAU Circ., 4614Kirshner, R. P., Leibundgut, B., & Horine, E. 1990, IAU Circ., 4953Kurucz, R. L. 1993, CD-ROM 1, Atomic Data for opacity Calculations

(Cambridge: Smithsonian Astrophysical Observatory)Lewis, J. R., Walton, N. A., Meikle, W. P. S., et al. 1994, MNRAS,

266, 27Lucy, L. B. 1991, ApJ, 383, 308Matheson, T., Filippenko, A. V., Ho, L. C., Barth, A. J., & Leonard,

D. C. 2000, AJ, 120, 1499Matheson, T., Filippenko, A. V., Li, W., Leonard, D. C., & Shields,

J. C. 2001, AJ, 121, 1648

Millard, J., Branch, D., Baron, E., et al. 1999, AJ, 527, 756Nakano, S., Aoki, M., Kushida, R., et al. 1996, IAU Circ., 6454Nomoto, K., Suzuki, T., Shigeyama, T., et al. 1993, Nature, 364, 507Patat, F., Chugai, N., & Mazzali, P. 1995, A&A, 299, 715Patat, F., Cappellaro, E., Danziger, J., et al. 2001, ApJ, 555, 900Phillips, M. M., Heathcote, S. R., Hamuy, M., & Navarrete, M. 1988,

AJ, 95, 1087Qiu, Y., Li, W., Qiao, Q., & Hu, J. 1999, AJ, 117, 736Schlegel, E. M., & Kirshner, R. P. 1989, AJ, 98, 577Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, 500, 525Shigeyama, T., Suzuki, T., Kumagai, S., et al. 1994, AJ, 420, 341Stritzinger, M., Hamuy, M., Suntzeff, N. B., et al. 2002, AJ, 124, 2100Swartz, D. A., Clocchiatti, A., Benjamin, R., Lester, D. F., & Wheeler,

J. C. 1993, Nature, 365, 232Utrobin, V. P., & Chugai, N. N. 2002, AstL, 28, 386Wheeler, J. C., Harkness, R. P., Clocchiatti, A., et al. 1994, ApJ, 436,

135Woosley, S. E., Eastman, R. G., Weaver, T. A., & Pinto, P. A. 1994,

AJ, 429, 300Woosley, S. E., Langer, N., & Weaver, T. A. 1995, ApJ, 448, 315

hydrogen and helium traces in type ib-c supernovae

Documents